Tag Archives: managed ddos protection uk

DDOS Attacks on Healthcare Providers: 6 Weak Points

DDOS Attacks on Healthcare Providers: 6 Weak Points

Picture the first Tuesday clinic of the month at a district general hospital: twenty-eight outpatient slots booked through a third-party portal, a pathology feed arriving over SFTP from a regional lab, community nurses logging in through a VPN concentrator that was sized in 2019, and a website that quietly stopped being the trust’s problem the day the original web agency took over the domain. Then the portal stops answering. DDoS attacks on healthcare providers rarely succeed because the traffic volume was extraordinary; they succeed because one of those components had no owner, no mitigation and nobody awake to act on it.

This piece does not rehearse sector statistics. It names the six places UK health estates actually break during a denial-of-service event, and what to put in a contract so they do not break twice.

Why health estates fall over at volumes other sectors absorb

A bank runs its own autonomous system, its own IP space and a security operations centre with a rota. A trust runs a federation. The main website might sit with a regional NHS host, the patient portal with a clinical software vendor, e-referral with another, video consultation with a third, the integration engine on a virtual machine in a shared data centre, and the authoritative DNS wherever it was parked a decade ago. Six owners, six contracts, six different answers to the question of who is allowed to phone a mitigation provider at 02:00.

The damaging outcome in a clinical setting is not data theft. It is unavailability. A booking engine that is unreachable for ninety minutes produces paper fallback, phone triage, a clinic list worked from printouts, and a rebooking backlog that is still being cleared weeks later. Ambulance diversion is the extreme case, but the ordinary case is quietly expensive: consultant time burned on a stalled list, administrative overtime, and appointment slots that cannot be recovered because the capacity simply does not exist.

Motive shapes the target, not the size of the pipe

Hacktivist pressure campaigns pick targets for visibility, which means public-facing portals and anything with a recognisable name. The US Department of Health and Human Services’ Health Sector Cybersecurity Coordination Center (HC3) published an analyst note in January 2023 on the pro-Russian group KillNet targeting the health sector with denial-of-service attacks, and pro-Russian groups have since claimed attacks against European public sector sites. Ransom-motivated actors behave differently: they time the disruption around a known pressure point, such as a winter surge or the first working day after a bank holiday, and they test which surface hurts most. Reading the motive early tells you whether to expect a short noisy burst or a sustained campaign, which is the practical value of understanding what hacktivist attack motives predict.

Weak point 1: authoritative DNS that nobody in the trust owns

Ask an information governance lead who manages the trust’s authoritative nameservers and the answer is often a pause. The zone is frequently still with the agency that built the original site, or inside a legacy hosting account whose billing contact left in 2018. Domain Name System (DNS) is the layer most reliably left unmanaged in health estates, and it is the one with the widest blast radius.

A query flood against a single authoritative provider does not take down one website. It takes down every name in that zone at once: the patient portal, staff email routing, the identity provider used for single sign-on, and any API hostname your suppliers resolve. Web application firewalls and reverse proxies are irrelevant at that point, because nothing ever resolves far enough to reach them.

Three practical steps. Inventory every zone the organisation depends on, including the ones registered by departments and charities under your name. Check whether you have genuinely diverse authoritative nameservers (two providers, separate anycast networks, not two hostnames on one platform). Set time-to-live values you can actually work with during an incident, because a 24-hour TTL means your emergency change takes a day to propagate. The detail of provider selection is covered in this buyer’s guide to authoritative DNS protection.

Questions worth putting to your DNS host in writing: what query-per-second capacity do you provision per customer, how many points of presence carry our zone, and who answers the phone at 02:00 on Boxing Day?

Weak point 2: origin IP leakage behind the patient portal

Proxy bypass is the most common way a protected site falls over, and it is almost never a capacity problem. The attacker has simply found a route that goes around the proxy entirely.

Health estates leak origins in predictable places: an old A record for a decommissioned booking subdomain that still points at the real server, a UAT or staging environment on the same subnet, a mail host sharing the range, a certificate transparency log entry naming an internal hostname, or a PDF and imaging endpoint serving large files straight from the origin because someone excluded it from the proxy for performance. Attackers enumerate those first. Appointment booking and repeat prescription ordering are the endpoints they probe hardest, because both are expensive per request on the back end.

The fix is unglamorous. Lock the origin firewall to the proxy’s published address ranges, deny everything else, then verify it from outside your network rather than assuming the rule works. Re-verify after every migration. Application-layer protection is only as good as the path discipline underneath it, which is the recurring theme of these application layer protection gaps.

Weak point 3: clinical SaaS and integration layers outside your contract

Most clinical availability depends on networks the trust does not control. Booking engines, e-referral, remote patient monitoring, pathology result delivery, prescription services and video consultation platforms all sit on someone else’s infrastructure, often in shared tenancy where an attack aimed at a neighbour can degrade your service.

Protection so belongs in supplier contracts, not only in your own hosting arrangement. Insist on four things in writing: the named mitigation provider and the underlying network doing the scrubbing; a stated time to mitigate with the conditions under which the clock starts; a notification duty within a defined window when the supplier is under attack; and a written post-attack report with a vector breakdown.

Then do the mapping exercise. Take each clinical pathway and ask what happens if that one supplier is unreachable for two hours on a weekday morning. Some pathways degrade gracefully. Others stop. The ones that stop are the ones to argue about at contract renewal.

Weak point 4: remote clinician access and the network edge

VPN concentrators, remote desktop gateways and session brokers are low-capacity choke points sitting in front of high-value work. A state exhaustion attack fills the session table or the TLS handshake queue while using a fraction of the link’s bandwidth. Your monitoring shows the circuit at 12 per cent utilisation and the clinicians cannot log in. That combination causes more wasted triage time than any volumetric flood.

Reverse proxy protection does nothing here, because none of this is HTTP traffic you can route through a content delivery network. If your organisation holds its own IP space and autonomous system number, Border Gateway Protocol (BGP) based scrubbing covers the whole range, including SFTP result feeds, HL7 interfaces and API endpoints. If it does not, you need per-service protection from whoever owns the addresses, and you need to know that before an incident rather than during one.

Weak point 5: on-demand mitigation that misses clinical timeframes

On-demand mitigation sounds economical. Detection, then a BGP announcement, then route convergence, then traffic actually arriving at the scrubbing centre. Each step costs real minutes, and minutes map directly onto cancelled slots during a running clinic.

Read the service level agreement for what triggers the clock. A “15 second time to mitigate”usually assumes always-on traffic baselining, pre-agreed thresholds and signed-off authorisation; without those, the clock starts when a human on your side raises a ticket and gets authenticated. Ask who in your organisation is authorised to trigger mitigation out of hours, and check that person exists on an August bank holiday weekend. Many escalation matrices name a role rather than a rota.

Diversion has trade-offs worth stating openly: added latency on every clinical session, asymmetric routing that can break stateful appliances, and services (voice, some VPN tunnels, IP-allowlisted supplier feeds) that fail when source addresses change. Test those before you need them, not at 09:15 on a Monday.

Weak point 6: a ‘managed’ contract that is really a portal login

The dividing line is simple. Managed means onboarding and rule tuning against your real traffic, a tested runbook, a named escalation path and an engineer empowered to change policy during the attack without waiting for your ticket. If nobody on the provider’s side can act without your sign-off, you have bought monitoring, not management, and out of hours that means nobody acts at all.

In public sector procurement the resale chain matters. The logo on the quote is often not the network doing the scrubbing; capacity, peering and support hours frequently belong to a provider two steps down the chain. Ask who operates the scrubbing centres, where they are, and whether support is in-house or subcontracted overnight. Then judge on evidence: sample post-attack reports with real vector breakdowns, references from organisations of comparable size, and a straight answer on how false positives are handled.

Healthcare tuning has one rule generic providers get wrong. Never rate-limit a clinician into a locked-out state. Hospital networks present hundreds of users behind a handful of NAT addresses, kiosk browsers carry odd user agents, and automated feeds (results, device telemetry, interoperability polling) look exactly like scripted abuse to a default ruleset. Allowlist them deliberately, by source and by behaviour, before go-live. General principles on how denial-of-service attacks work and how they are mitigated are a useful grounding for the governance papers that accompany this sort of procurement.

Building the business case and surviving the first hour

Price the downtime before you price the protection. Take one cancelled outpatient list: the lost slots, consultant and nursing time spent on paper fallback, administrative overtime, out-of-hours recovery, and the rebooking backlog that spreads across the following weeks. Put a figure against it using your own costing data, then set that beside an annual protection contract. That comparison persuades finance directors; fear does not.

UK managed protection is usually priced on a committed bandwidth or protected-IP basis with a monthly fee, plus onboarding, and often with burst or overage charges once attack traffic exceeds a threshold. The line items buyers forget at first invoice are additional protected subnets, extra portal users, out-of-scope incident response hours and the cost of a second mitigation point for DNS. The breakdown in this guide to what UK DDoS protection actually costs is a reasonable starting point for a budget paper.

For the first hour, work in sequence. Confirm it is an attack rather than a failed deployment or an upstream fault, using the approach in this piece on telling a DDoS apart from an ordinary outage. Check three surfaces in order: DNS resolution, the proxy or edge, then the origin and the remote access concentrators. Notify affected clinical suppliers and ask them to confirm their own status. Capture evidence as you go (packet samples, flow records, timestamps, portal screenshots), because reconstruction afterwards is always worse.

Governance follows. Inform the on-call director and the information governance lead, log the incident against the NHS Data Security and Protection Toolkit processes your organisation already uses, and consider obligations under the Network and Information Systems Regulations 2018 if your organisation is designated an operator of essential services. UK GDPR Article 32 expects availability and resilience of processing systems, so a sustained outage is a security matter, not only an operational one. Denial-of-service attacks are offences under the Computer Misuse Act 1990, which is the basis for reporting to Action Fraud and, for significant incidents, to the National Cyber Security Centre.

Frequently Asked Questions

Why are healthcare providers targeted by DDoS attacks?

Visibility and urgency. Hacktivist groups choose targets whose disruption gets reported, and health services qualify; HC3 in the United States issued an analyst note in January 2023 about pro-Russian actors targeting the sector. Extortion-motivated attackers pick healthcare because the tolerance for downtime is close to zero, which increases the chance of a fast payment or a fast concession.

Can a DDoS attack on a hospital affect patient care directly?

Yes, indirectly but materially. If booking, e-referral, results delivery or remote clinician access is unreachable, staff revert to paper and telephone, lists slow down and appointments get cancelled. The clinical risk sits mostly in delayed care and the backlog, rather than in any loss of patient data.

Does a patient portal behind a reverse proxy still need network-level protection?

Usually, yes. A proxy only protects traffic that goes through it, so a reachable origin, an exposed UAT host or a mail server on the same range reopens the direct path. Non-HTTP services (VPN, SFTP feeds, HL7 interfaces) are not covered by a web proxy at all.

How much does managed DDoS protection cost for an NHS trust or private hospital group?

Pricing varies by committed clean bandwidth, number of protected IP ranges, always-on versus on-demand, and support hours, so there is no single figure worth quoting. Build the comparison from your own downtime cost per cancelled clinic list, and ask every bidder to quote onboarding, overage and out-of-scope incident hours separately.

What should we do in the first hour of a suspected DDoS attack on a clinical system?

Confirm it is an attack rather than a deployment or upstream failure, then check DNS, the edge and the origin in that order. Notify your mitigation provider and any affected clinical suppliers, start the clinical fallback process, and record evidence with timestamps for reporting to Action Fraud and the NCSC and for any later contractual or legal action.



DDOS Scrubbing Capacity: How to Judge Tbps Claims

DDOS Scrubbing Capacity: How to Judge Tbps Claims

Open almost any mitigation provider’s data sheet and the DDoS scrubbing capacity figure sits near the top, in bold, usually followed by a unit that ends arguments: 10 Tbps, 30 Tbps, 200 Tbps and rising. The number is there to close the conversation. It should open it, because the figure on the page and the capacity that will actually stand between your prefix and a botnet on a Tuesday afternoon are rarely the same thing.

Scrubbing capacity means the volume of attack traffic a provider can ingest, inspect and filter before passing clean traffic on to you. Simple enough. The difficulty is that vendors measure it in at least three incompatible ways, and the one they publish is almost always the most flattering.

What a scrubbing capacity figure is actually measuring

Three distinct numbers get printed under the same label.

Total network capacity is everything the provider’s backbone can carry, including transit and content delivery traffic that has nothing to do with mitigation. Aggregate mitigation capacity is the sum of filtering capability across every scrubbing location they operate. Per-centre capacity is what a single site can ingest and clean. Only the third one is about you.

At the time of writing, published figures span two orders of magnitude in framing. Some platforms quote well over 100 Tbps spread across several hundred points of presence; others quote around 30 Tbps across roughly two dozen dedicated scrubbing centres. Do the division before you do anything else. A headline that divides into a few hundred gigabits per location is telling you something very different from one that divides into several terabits, and the fewer, larger model is often the stronger one for a single-region customer. Capacity is never distributed evenly anyway, which is exactly why the next question matters: what is the ingest and mitigation capacity of the specific centre my traffic is diverted to, and what happens when that centre is the one under pressure?

Anycast changes the question

On an anycast platform, a volumetric flood is absorbed wherever the sources are, which is genuinely useful against a globally distributed botnet. It is less useful when the attack sources are concentrated in Europe and your diversion lands in one or two European sites. Aggregate capacity across hundreds of global locations is an architecture statement. It is not a promise about your prefix.

Inside the platform there are also per-customer ceilings that have nothing to do with the network: a clean-traffic commit, a cap on how many protected prefixes or hostnames you may onboard, and in appliance-backed or licensed models a throughput entitlement written into the order form. A 100 Tbps platform will happily enforce a 500 Mbps licence against you.

Where UK traffic lands, and why the nearest centre decides the outcome

For most UK buyers the relevant ingest points are London and Slough, and the relevant peering is at LINX and LONAP. Ask which facility your prefix would be diverted into, what that site’s ingest capacity is, and where it fails over. The answer to the third question is frequently Amsterdam or Frankfurt.

That failover is not neutral. London to Frankfurt is in the region of ten to fifteen milliseconds of round-trip time on a good path, and a checkout flow or a chatty API that makes twenty or thirty sequential calls will show it. Sessions that were comfortable become marginal. Timeouts tuned for in-country latency start firing. You mitigated the attack and still lost conversions, which is a conversation worth having before you sign rather than during an incident.

Data centre DDoS protection sold at the rack deserves particular scepticism. The colocation provider may advertise protection, but the filtering usually happens at an upstream transit provider’s edge or a third-party scrubbing partner, under a contract you have never read, with a threshold you have never been told. Ask who owns the mitigation, not who sold it.

Bits per second is the wrong unit for most real attacks

This is the part buyers underestimate. Forwarding hardware fails on packet rate long before it fails on bandwidth. At minimum Ethernet frame size, a 10 Gbps link carries roughly 14.9 million packets per second; 100 Gbps of 64-byte packets is close to 149 Mpps. Routers, firewalls and intrusion-prevention appliances run out of forwarding capacity and state-table headroom at those rates while the bandwidth graph still looks calm. A flood that never approaches an advertised terabit ceiling can still flatten an edge device.

Carpet bombing exploits the other half of the problem: detection thresholds. Spread a few hundred megabits per second across the 1,024 addresses of a /22 and you have tens of gigabits arriving at the network while no single IP crosses a per-host trigger set at, say, 1 Gbps. Nothing is detected. Nothing is diverted. Detection configuration, in that scenario, matters far more than the headline on the data sheet, and the same arithmetic applies to the terabit-scale attacks that make the headlines.

Layer 7 is a different unit again. Application floods are measured in requests per second and new connections per second, and TLS handshakes are deliberately asymmetric: cheap for the attacker, expensive for your termination point. Gigabits are irrelevant there. The relevant figures are proxy request capacity and policy quality, which is a separate discipline covered in our breakdown of where application layer defences actually fail.

So ask for three numbers, not one: Tbps, Mpps and concurrent sessions or requests per second, per scrubbing centre. Providers who track their own platform properly can answer. The ones who cannot are telling you something.

The contract terms that quietly resize your capacity

A mitigation can succeed technically and still cost you money or availability, because the commercial terms cap what the platform is allowed to deliver.

  • Clean-traffic commit and overage. You commit to a volume of clean traffic, often billed at 95th percentile. Mitigation works, legitimate traffic keeps flowing, and the invoice arrives with overage at a rate nobody modelled. Check whether attack traffic is excluded from the measurement, in writing.
  • Black hole thresholds. Lower tiers at hosting and cloud providers frequently protect up to a stated rate and null-route the address above it, usually by announcing it with the BLACKHOLE community defined in RFC 7999. The attacker gets exactly what they wanted, delivered by your own provider. Find the threshold in the plan documentation before you buy, and find out how long the null route stays in place.
  • Attack-size caps and fair use. “Unlimited mitigation”often carries a clause permitting renegotiation after repeated incidents, or a cap on single-attack volume buried in an appendix.
  • SLA credits. These are written against time-to-mitigate and availability, never against capacity. No provider refunds you for failing to deliver 30 Tbps.

Those terms, not the engineering, are where most unpleasant surprises live. We have gone through the pricing models in more detail in our guide to what UK DDoS protection really costs.

Whose capacity are you buying?

A large share of UK managed DDoS protection offers are resold. Hosting companies, managed service providers and smaller security firms front someone else’s scrubbing network under their own brand, which is not inherently a problem; it becomes one when nobody in the chain can change a policy during an incident.

Tracing the real network is straightforward. Look up the announcing AS for the protected prefix during and outside mitigation using RIPEstat or any public looking glass. Check the CNAME chain and the name servers. Read the vendor name in the footer of the attack report you are shown. Then ask directly: whose AS announces my prefix when I am under attack, and who has authority to change filtering policy on it?

The dividing line is simple. If a human on the provider’s side cannot adjust policy at 02:00 without hunting for your sign-off, you have bought monitoring, not management. Identical underlying capacity produces very different outcomes depending on who is tuning the signatures, writing the rate limits and making the diversion call, which is also why a clear method for separating an attack from an ordinary outage belongs in the runbook on your side too.

Capacity across all three surfaces, including the one left out

Protection is usually sold for one surface and assumed for three.

Network and transport. BGP diversion to a scrubbing centre, returned over GRE or a cross-connect. This is the layer the Tbps figure describes, and the only one it describes.

Application. Reverse proxy capacity in requests per second, plus WAF rule quality and bot classification. Raw bandwidth tells you nothing here.

Authoritative DNS. Routinely outside the protection contract. A site can sit behind a very large scrubbing platform while its zone is served by a provider with a handful of anycast nodes, making name resolution the cheapest thing in the estate to knock over. Our buyer’s guide to authoritative DNS protection covers what to ask for there.

And then there is origin exposure, which defeats capacity entirely. Proxy bypass is far more often caused by a leaked origin address than by an attack that genuinely outsizes the platform. The usual culprits: an MX record pointing at the same host, an old A record on a forgotten subdomain, historical DNS data, certificate transparency logs listing every hostname you have ever requested a certificate for, and an admin or SSH endpoint reachable on the origin IP. Audit those first. Then restrict the origin to accept traffic only from your provider’s published ranges. Unlimited scrubbing capacity in front of a directly reachable server protects nothing.

Sizing from evidence rather than data sheets

Start with your own numbers, not the vendor’s. Take peak legitimate throughput, peak packet rate and peak requests per second from the last twelve months, add realistic growth headroom, and work out what an hour of downtime costs in revenue, transaction value, contractual penalties or statutory deadlines missed. Public sector and regulated services often find the deadline matters more than the revenue. The NCSC’s denial of service guidance makes the same underlying point: understand what normal looks like for your service before you try to specify protection for it.

Then decide between always-on and on-demand. On-demand is cheaper and introduces a diversion window governed by detection time, BGP convergence and, for DNS-based redirection, your record TTL. Always-on removes that window and costs more every month. Compare the monthly difference against the cost of the diversion window, not against an abstract preference for one architecture.

Pre-contract checklist

  1. Named ingest locations serving UK prefixes, with per-centre capacity and the designated failover site.
  2. Published Mpps and concurrent-session or requests-per-second limits, not just Tbps.
  3. Per-IP and per-prefix detection thresholds, and how carpet bombing across a /22 is detected.
  4. Any black hole or null-route threshold in your tier, in writing, with the removal process.
  5. Clean-traffic commit, overage rate and whether attack traffic is excluded from billing.
  6. Time-to-mitigate definition, measured from what event, with the SLA credit mechanism.
  7. A redacted post-incident report from a real mitigation, with the vector breakdown.
  8. The escalation matrix: named contacts, out-of-hours route and who authorises diversion.
  9. The announcing AS during mitigation, confirming whose network you are buying.

Run that against every quote. A provider with genuine DDoS scrubbing capacity behind the marketing will answer all nine in a single call; the ones who keep returning to the headline terabit figure are telling you where their confidence actually sits.

Frequently Asked Questions

What does DDoS scrubbing capacity actually mean?

It is the volume of attack traffic a provider can ingest, inspect and filter before forwarding clean traffic to you. The complication is that the published figure may describe total backbone capacity, mitigation capacity summed across every location, or the capacity of one scrubbing centre. Only the last of those relates to what protects your prefix.

Is a higher Tbps figure always better protection?

No. A larger aggregate spread thinly across hundreds of locations can leave less capacity at the specific site serving your traffic than a smaller platform with a few large centres. Packet rate, detection configuration, policy tuning and contract terms all decide outcomes that bandwidth alone does not.

How much scrubbing capacity does a typical UK business need?

Size it against your own peak legitimate throughput and packet rate with growth headroom, rather than against the largest attack in the news. For most organisations the binding constraints are the per-customer commit, the detection thresholds and whether the origin is reachable directly, not the platform ceiling.

What is a black hole threshold and how does it limit mitigation?

It is a rate above which your provider stops filtering and simply discards all traffic to the targeted address, usually by announcing it with the BLACKHOLE community from RFC 7999. The attack stops reaching your network and so does everything else. Entry-level hosting and cloud plans commonly include one, so check the documented threshold and the removal procedure before buying.

How can I check whose scrubbing network my provider is really using?

Look up the announcing autonomous system for your prefix on a public looking glass or RIPEstat, follow the CNAME and name server chain, and read the vendor name on any attack report you are given. Then ask the provider directly who announces your prefix during mitigation and who is authorised to change policy out of hours.



Law Firm Cyber Attack Protection: Beyond Phishing

Law Firm Cyber Attack Protection: Beyond Phishing

Law firm cyber attack protection, as it is sold and as it is bought, nearly always means email: a filtering gateway, phishing simulations twice a year, a mandate fraud warning on the completion statement, and a cyber policy with a conveyancing fraud sub-limit. That half of the problem has had a decade of attention, and most UK firms have genuinely improved. The other half has had almost none. Nobody at renewal asks what happens when the client portal, the document exchange and the firm’s authoritative DNS all stop answering at three o’clock on a Friday afternoon.

That is the gap this piece is about: availability, how it fails, and how to buy and verify protection for it without taking a vendor’s word for anything.

What attackers actually target in a law firm, and the half nobody covers

The confidentiality attacks are familiar. Business email compromise, invoice and mandate redirection on completion funds, credential theft against Microsoft 365, ransomware that encrypts a file share and the backups sitting on the same domain. Firms have bought tooling and training against all of it, and insurers have pushed hard in the same direction.

Availability attacks sit outside that conversation entirely. They include volumetric floods against the public website; request-level attacks aimed at the login flow and search function of a client portal or document exchange; attacks on the authoritative nameservers that publish every hostname the firm uses; and extortion emails demanding payment in cryptocurrency to stop, sometimes preceded by a short demonstration burst against the marketing site.

Legal is an attractive target for both motives, and for different reasons. Work is deadline-bound, so an outage has a hard commercial edge that an attacker can time. Clients in litigation and corporate transactions are acutely sensitive to any sign that the firm is unstable. And high-value matters are often publicly visible, from planning objections to M&A announcements, which gives an aggrieved party a clear target and a clear window. One UK practice found its systems disrupted badly enough by a flood of inbound mail that it rebuilt its web security afterwards, which is a reminder that the trigger is not always sophisticated and rarely announces itself in advance.

Mapping your own exposure in an afternoon

Before anyone looks at a quote, build four columns. Every client-facing hostname the firm owns, including the ones marketing created and forgot; who hosts each one; who runs the DNS for the zone; and the name and mobile number of the person you would ring about it at two in the morning. Most firms cannot complete column four for at least one row. That blank is the finding.

The three surfaces law firm cyber attack protection has to cover

Network and transport floods

These are the attacks discussed in gigabits per second, occasionally terabits. A decent hosting provider or ISP absorbs the small end without telling you. What they will not do is carry a sustained attack aimed at your single public IP address, because at a certain point the economically rational move is to null-route you to protect everyone else on that segment. Blackholing is not mitigation. It is the provider agreeing with the attacker.

Application-layer attacks

This is where portals actually fall over. A few thousand well-chosen requests per second against a document search, a login endpoint that does a bcrypt hash on every attempt, or a matter-lookup page that runs an unindexed query, will exhaust the database long before anything registers on a bandwidth graph. Protection sized purely on gigabits of clean traffic will sail through a volumetric flood and still let the conveyancing portal die quietly. Understanding the gaps in application-layer defences matters more for a law firm than headline scrubbing capacity, because the valuable systems are all request-driven.

Authoritative DNS, the layer nobody has reviewed

Ask who controls the firm’s nameservers. In a worrying number of practices the answer is a web agency’s registrar account, sometimes a former IT supplier’s, on a single DNS provider, with no secondary nameservers and no monitoring. Lose that and nothing else in the stack matters: the portal, webmail, the marketing site and any MX-dependent mail flow all stop resolving at the same moment, and the shiny proxy in front of the website never sees a packet. Redundancy here is cheap and badly under-bought; a second authoritative DNS provider is usually the highest-value hour of work available to a firm this week.

Third-party dependencies you do not control

Practice management SaaS, e-signature platforms, hosted document portals and card payment pages all inherit someone else’s protection posture. Your Lexcel file says you have business continuity; your supplier’s contract may say they will use reasonable endeavours. Those are not the same sentence. Put the question in writing to each supplier at renewal and keep the answer.

Why the proxy gets bypassed: origin IP leakage, not attack size

When a protected site goes down, the cause is far more often that the attacker found the origin than that they out-muscled the scrubbing network. The firm’s real server address leaks through routes nobody audits: the MX record, SPF entries listing the mail server, historical DNS records archived by passive DNS services, TLS certificate transparency logs that publish every hostname you have ever certificated, a webmail or VPN endpoint sitting on the same /24 as the web server, or a staging site at new.firmname.co.uk that resolves straight to the origin and has done since 2021.

The test is one question, and it is a yes or no. Does the origin accept TCP connections on 80 and 443 from anything other than the protection provider’s published IP ranges? If yes, the protection in front of the site is decorative. Three fixes, in order: allow-list the provider’s ranges at the host firewall or security group and drop everything else; validate the host header and a shared secret header at the web server so direct hits are rejected; and rotate the origin IP address after onboarding, because the old one is already in a passive DNS database somewhere.

Attack volume is the part vendors like to talk about. Reachability is the part that decides the outcome.

Extortion and the first hour

Triage before escalation

Not every Friday outage is an attack. A bad deployment, an expired certificate, a hosting provider’s own incident and a legitimate traffic spike from a press mention all look identical from reception. Check from outside the network, not from a machine on the office LAN; compare the web server’s request logs against the load balancer’s; look at whether the source addresses are geographically scattered and hitting one expensive URL repeatedly. Having a short, written routine to tell a DDoS from an ordinary outage saves the twenty minutes that usually get burned arguing.

If a ransom demand arrives

Paying achieves nothing reliable. Payment marks the firm as responsive to pressure, and the group that sent the email may not be the group capable of stopping the traffic. Preserve instead: the email with full headers, the exact timestamps of the first and peak traffic, sample requests and source data from the logs, and the provider’s post-attack report with its vector breakdown. That evidence is what makes any later legal or law enforcement route viable rather than theoretical.

Reporting duties

Four obligations sit in different places and a firm should know which apply before the day arrives:

  • The SRA Standards and Regulations require prompt reporting of serious breaches of the regulatory arrangements, which can include an incident that materially affects the firm’s ability to act for clients;
  • UK GDPR requires notification to the Information Commissioner’s Office within 72 hours of becoming aware of a personal data breach where there is a risk to people’s rights and freedoms, and the ICO is explicit that a loss of availability counts as a breach, not just unauthorised disclosure;
  • Clients on live matters need telling, in plain terms, where a deadline or completion is affected, and that message is better coming from the partner than from an estate agent;
  • Action Fraud takes the crime report, and the National Cyber Security Centre should be notified where the incident is significant.

Denial of service is a criminal offence in the UK under section 3 of the Computer Misuse Act 1990, which covers unauthorised acts intended to impair the operation of a computer. Useful to know, and worth stating to staff who assume this is a grey area. Realistically it does not shorten your outage by a second, and attribution is slow, so treat the criminal route as evidence preservation rather than recovery.

Buying protection: deployment models, managed meaning, and the cost case

Three models, and the choice is mostly determined by what the firm owns. Reverse proxy suits the common case: one marketing site, a client portal, a document exchange, all behind DNS you can repoint, with no IP space of your own. BGP scrubbing is for firms announcing their own prefix, typically those running an on-premises data room or legacy case system, and it protects everything in the range including mail and VPN. Hosting-level filtering is the cheapest and the weakest; it tends to stop at the volumetric tier and leaves the application layer to you.

On always-on versus on-demand: on-demand diversion is cheaper and adds a detection window plus a BGP announcement and propagation window before mitigation begins. For a firm whose work turns on 16:00 filing cut-offs and completion days, those minutes are a commercial decision, not a technical footnote. Decide now who is authorised to trigger a diversion at 23:00 on a Sunday without a partner’s sign-off, and write the name in the runbook.

What “managed”has to mean in the contract

Managed is a word, not a service level. The dividing line is simple. If a human on the provider’s side is not empowered to change a rate limit or a WAF rule at 16:40 on a Friday while the conveyancing portal is being hammered, without waiting for someone at the firm to raise a ticket, you have bought monitoring, not management. Get four things in writing: the onboarding and rule tuning done before the first attack rather than during it; named escalation contacts on both sides with out-of-hours numbers; explicit authority to change policy mid-incident; and a written post-attack report with timestamps, vectors and actions taken.

The business case, built from downtime rather than fear

Do the arithmetic openly. Take the fee earners who would be unable to work for an afternoon, multiply by the hours lost and by your own charge-out rates, and you have the cost of that single incident in lost recoverable time, before you count a missed completion, an abortive lender drawdown or the partner hours spent on client calls. Compare that number to an annual managed protection quote. Do not compare quotes to each other, which is how firms end up buying the cheapest version of the wrong thing. The structure of UK pricing varies by model, so check current figures with providers directly rather than relying on anything published months ago.

Judging providers on evidence rather than sales decks

Much of what is marketed in the UK as DDoS-protected hosting resells someone else’s scrubbing network. That is not disqualifying, but you need to know it. Ask whose capacity you are buying, which points of presence serve UK traffic (London is not a complete answer if the only nodes are in Frankfurt and Amsterdam), and who physically answers the phone at three in the morning, because that determines both the real capacity and who is accountable mid-incident.

Then press on the SLA. What does it pay out, what triggers it, how is time-to-mitigate measured, and who measures it? A service credit worth one month of fees against a lost afternoon of fee earner time is a gesture, not a remedy. And ask for a redacted report from a real incident. Report quality tells you more about the operations team than any capacity figure on a slide: whether they identified vectors, when a human intervened, and what they changed.

Run this short list alongside Cyber Essentials, Lexcel and the insurer’s questionnaire. Note that Cyber Essentials and Cyber Essentials Plus, for all their value, assess firewalls, secure configuration, access control, malware protection and update management. None of those five controls test whether your site stays reachable under attack. Availability is the box nobody ticks because nobody asks, which is precisely why it is still the weak point in most firms’ answers. Effective law firm cyber attack protection closes it by naming the surfaces, locking the origin, and putting a human with authority on the other end of the phone before the first demand email arrives. Background reading on how these attacks are structured and resourced is collected across DDoSInfo, and the case of the firm that hardened its web security after an attack is a reasonable place to start with partners who need convincing.

Frequently Asked Questions

What does law firm cyber attack protection need to cover beyond email security?

Availability of every client-facing system: the public website, client login portals, document exchange, and the authoritative DNS that publishes all of them. Email controls address confidentiality and fraud; they do nothing if the portal is unreachable on a completion day. Treat network floods, application-layer request attacks and DNS as three separate surfaces with separate owners.

Can a small law firm’s website really be taken offline by a DDoS attack?

Yes, and it rarely takes a large attack. A booter service costs very little and a few thousand requests per second aimed at a login page or document search can exhaust the database behind it. Smaller firms are often more exposed because their site and portal share one modest origin server with no filtering in front of it.

What should a firm do if it receives a ransom demand threatening to take its site down?

Do not pay and do not reply. Preserve the email with full headers, confirm whether an attack is actually in progress, notify your protection provider and hosting supplier, and report to Action Fraud and, if the incident is significant, the NCSC. Payment marks the firm as responsive to pressure and offers no reliable guarantee the traffic stops.

Does a law firm have to report a denial of service attack to the SRA or the ICO?

It depends on effect, not on the label. The SRA Standards and Regulations require prompt reporting of serious breaches, which can include an incident that materially affects the firm’s ability to act for clients. Separately, the ICO treats loss of availability of personal data as a personal data breach, so a 72-hour notification may be required where there is a risk to individuals. Take advice from the COLP rather than deciding on the day.

Is always-on or on-demand mitigation better for a firm with court filing deadlines?

Always-on, in most cases. On-demand diversion is cheaper but adds detection, announcement and propagation time before mitigation starts, and those minutes land badly against a filing cut-off or a lender’s close of business. If budget forces on-demand, agree the out-of-hours authorisation route in advance and name the person who can trigger it without a partner’s approval.

How much should a UK firm expect to spend on managed DDoS protection?

Pricing varies by model: reverse proxy plans are usually billed per protected hostname or per volume of clean traffic, while BGP scrubbing is priced against committed capacity and the size of the prefix. Get current quotes directly, since published figures date quickly. The useful comparison is not quote against quote but quote against your own downtime cost, calculated from fee earner hours lost in a single three-hour outage.



DDOS Protection Cost UK: What You Really Pay

DDOS Protection Cost UK: What You Really Pay

Ask five UK providers to price protection for the same ecommerce site and the quotes will range from a small monthly subscription to an annual contract many times larger, all of them described as DDoS protection. That spread is not a negotiating tactic. The DDoS protection cost UK buyers face varies by two orders of magnitude because the word “protection”is being used to sell at least three different products, bought by three different sorts of customer, with wildly different amounts of human engineering attached. Work out which product you are actually being quoted and the numbers stop looking random.

What follows is a cost anatomy from the buyer’s side: what each pricing model charges for, which line items stay invisible until the first attack, and how to turn downtime into a figure your finance director will sign.

Why DDoS protection quotes in the UK swing so wildly

Three markets hiding behind one search term

The first market is bundled protection: filtering included with a hosting plan, a content delivery network (CDN) tier or a cloud load balancer. Marginal cost near zero, tuning near zero.

The second is mid-market managed service: a monthly fee, a named provider, some human involvement, usually a reverse proxy or a small scrubbing arrangement in front of a handful of properties. This is where most UK mid-market buyers end up, and where the contract language matters most.

The third is enterprise Border Gateway Protocol (BGP) scrubbing with committed capacity: you announce your own /24, traffic is diverted to scrubbing centres, clean traffic comes back over GRE tunnels or a cross-connect. You are buying network capacity, a routing relationship and an engineering team. It has a high floor because all three cost money whether or not you are attacked.

What an entry-level plan genuinely covers

Entry-level proxy plans advertised in the UK market at low monthly prices are real products, not bait. They buy a share of a large anycast proxy estate, automated volumetric filtering, a rate-limiting engine and a dashboard. For a brochure site or a small SaaS front end, that is often proportionate.

What they do not buy is anyone’s attention. Best-effort means exactly that: your traffic is defended by the same automated policy as everyone else on the shared tier, and if a layer 7 attack slips through the generic rules, the rule changes are yours to write. If nobody on the provider’s side is authorised to change policy at 03:00 without your written sign-off, you have bought monitoring, not management.

Why G-Cloud day rates are quoting people, not bandwidth

Public sector buyers looking at the Crown Commercial Service’s Digital Marketplace will see mitigation services listed at a price per unit per day. Those unit prices confuse people who assume they are paying for gigabits. They are not. G-Cloud style day rates typically price consultancy, onboarding, incident response time and diversion support, in other words a human being for a day. Comparing a day rate against a monthly subscription is comparing a plumber’s call-out charge with a water bill.

The question that explains most of the gap

Several UK “providers”do not own a scrubbing centre. They resell capacity on a larger network, wrap it in a portal and a support contract, and add margin. Sometimes there are two resale layers. Each one is another party that must be woken up before a mitigation policy changes, and another margin on your invoice.

So before you compare headline prices, ask who owns the scrubbing centres, whose autonomous system number (ASN) the traffic is cleaned on, and what your escalation path looks like in the middle of the night. Two quotes that differ by 10x for the “same”protection are usually not describing the same distance between you and the engineer who tunes the filter.

The pricing models behind the number on your quote

Per protected property or per protected IP. A flat monthly fee covering a defined number of domains, hostnames or IP addresses. The trap is counting creep. Staging, a customer portal, an API endpoint, a mail relay and a second data centre each add a chargeable unit, and the tidy monthly figure can multiply by the time the estate is honestly documented. Count every public-facing address before signing, not after.

Clean traffic bands with overage. The contract commits you to, say, 100 Mbps of clean traffic delivered, with an overage rate per Mbps or per GB beyond it. This is the single most common source of a surprise invoice after an attack, because attacks change traffic shape. Legitimate users refresh, bots retry, and some providers meter the traffic they forward after scrubbing rather than the flood they dropped. Read the metering definition, ask whether it is 95th percentile or peak, and ask what happens to billing during a declared incident.

Committed capacity and per-Gbps pricing. BGP scrubbing deals price a committed clean bandwidth tier plus the plumbing: GRE tunnel provisioning, or a cross-connect in a shared facility with its own monthly port fee. Budget for the network engineering time too. Announcing a /24 to a scrubbing provider, testing failover and documenting the withdrawal procedure is not a half-day job.

On-demand and emergency onboarding. On-demand looks cheaper on the monthly line because you pay for standby, not for constant traffic handling. The costs move elsewhere: diversion time while routes propagate, and in many contracts an emergency onboarding fee if you are attacked before you are provisioned. Compare the annual always-on premium against that fee plus the extra minutes of downtime, not against zero. Our breakdown of always-on versus on-demand DDoS protection works through where each model genuinely fits.

What each deployment model really costs once you add everything up

Reverse proxy or CDN-fronted protection

Lowest entry cost, priced on requests and bandwidth, quick to deploy with a DNS change. Its weakness is not capacity. It is origin leakage. If your server’s real IP address is still visible in historic DNS records, mail headers, TLS certificate transparency logs or an old subdomain, attackers route around the proxy entirely and hit the origin directly.

No pricing tier fixes an exposed origin. Budget for the remediation instead: new origin addressing, firewall allow-lists restricted to the provider’s published ranges, cleaning up stale A records and moving outbound mail off the web server. That work is usually a few days of engineering, and it is the difference between a paid proxy service that protects you and one that decorates your invoice.

BGP scrubbing and always-on routing

Higher floor, because you are buying capacity and a route rather than a shared proxy slot. It suits ASN holders, hosting companies, data centre operators and anyone protecting non-HTTP services such as game servers, VPN concentrators or VoIP. Costs to model: committed Gbps, tunnel or cross-connect fees, a /24 you actually control, and the internal staff time to run the diversion.

Hosting-level filtering

Bundled with a DDoS protected hosting package, near-zero marginal cost, and genuinely useful against crude volumetric floods. What you rarely get is tuning, a named escalation contact or a forensic report afterwards. If the provider’s answer during an incident is to null-route your IP to protect their other tenants, your site is down either way.

Whichever model you choose, the sticker price is not the total. Add web application firewall (WAF) licensing if it is not included, TLS termination and certificate management, log egress charges if you ship traffic logs into a SIEM, and internal staff hours. On a mid-market deployment those adders routinely equal the mitigation fee itself.

The line items buyers forget until the first invoice

Authoritative DNS. Left off the quote more often than any other component, then added later and billed separately by query volume. A site sitting behind a fully mitigated proxy is still unreachable if its name servers fold, and name servers are a soft target precisely because they are UDP-based and often hosted with a registrar nobody has thought about since 2019. Price network and transport layer, application layer and DNS as one budget line. Our buyer’s guide to authoritative DNS protection covers what to demand from that part of the stack, and the NCSC’s guidance on denial of service attacks is worth putting in front of anyone who thinks the web tier is the whole problem.

Attack reports and forensic exports. Sometimes a paid add-on, sometimes only available on enterprise tiers. Yet a vector breakdown with timestamps, source distribution and packet rates is the evidence you need for an insurer, a regulator, your own customers under an uptime clause, or any legal action following a DDoS attack. If the report costs extra, that is a line item, not a nicety.

What “managed”buys. The most expensive undefined word in a DDoS contract. Pin it down in writing: will the provider change rules mid-attack on their own authority, or does every tuning decision wait on a ticket and your approval? The approval model is cheaper monthly and far more expensive per incident, because the clock runs while your on-call engineer finds the portal password. Ask how many rule changes are included per year and what a change request costs beyond that.

SLA credits. Time-to-mitigate commitments are meaningful only if the measurement method is stated. There is a large practical difference between “60 seconds from detection”and “60 seconds from customer report”, and credits are almost always capped at a percentage of the monthly fee. A 100% credit on a £900 month does not cover a £20,000 trading outage. Treat credits as a signal of confidence, not as compensation.

Price your downtime before you price the protection

The business case is arithmetic, not fear. Build it from three components: revenue or transaction value per hour, incident staff cost, and contractual exposure to your own customers.

Take an illustrative mid-size UK retailer selling online. Spread across realistic trading hours, its annual online turnover gives an average revenue per hour, but peak Saturday afternoons in the run-up to Christmas might run several times that. A four-hour partial outage on a peak weekend, assuming half of attempted orders are lost rather than deferred, removes a meaningful slice of a trading weekend’s gross revenue. Add the people pulled onto an incident bridge for a day, plus an out-of-hours agency callout, and the total climbs again before anyone mentions refunds or reputational damage.

Against that, the annual cost of a managed DDoS protection UK contract is a fixed, known number. The break-even question becomes concrete: does your organisation expect more than roughly one significant outage a year? For an ecommerce operator during peak trading, a law firm running a client document portal under a professional obligation, a healthcare supplier with booking systems, or a B2B platform with liquidated damages in its uptime clauses, the answer is usually yes.

Resist the temptation to inflate. The Department for Science, Innovation and Technology publishes an average cost of the most disruptive breach in its annual Cyber Security Breaches Survey, and boards increasingly recognise that figure. It covers all breach types, skews low because it includes micro-businesses, and will be picked apart if you present it as a DDoS number. Quote the current edition by name and date if you use it, then rely on your own revenue-per-hour model for the actual case.

How to compare UK quotes on evidence rather than sales decks

Send every shortlisted provider the same ten questions and compare the answers side by side:

  1. Who owns the scrubbing centres, and are you a reseller of another network’s capacity?
  2. What is the total mitigation capacity, and what capacity is committed to me contractually?
  3. Is mitigation always-on or on-demand, and what is the measured time to divert?
  4. Can your engineers change mitigation policy mid-attack without my written approval?
  5. How is time-to-mitigate measured, from detection or from my report?
  6. What clean traffic is included, how is it metered, and what is the overage rate?
  7. How many protected IPs or hostnames are included, and what does each additional one cost?
  8. Is authoritative DNS protection included, and is it billed by query volume?
  9. Is a post-incident attack report with vector breakdown included at no extra charge?
  10. What is the named escalation path at 03:00 on a bank holiday, and what phone number do I call?

Then ask for something sales teams rarely offer: a redacted attack report from a real incident in the last twelve months. A capability slide proves nothing. A report showing attack start time, vectors, peak packet rate and the minute mitigation engaged tells you whether the operations team exists. If you want a structured starting point for evaluating vendors, our DDoS protection service comparison sets out how the main approaches differ in practice.

Terms worth negotiating: minimum term (push back on 24 months if you have not tested the service), a burst allowance that waives overage during a declared attack, waiver of emergency onboarding fees, and an exit clause that includes DNS TTL handover so you can leave without a self-inflicted outage.

Red flags, in order of seriousness: “unlimited”capacity claims, which no network can honour once you read the fair use clause; “fully managed”with no definition of who is authorised to act; mitigation SLAs with no stated measurement method; and any provider who will not name the network their traffic is cleaned on. One more, quieter one: a quote that covers only HTTP when half your revenue runs through an API or a DNS zone nobody has audited.

The honest summary of DDoS protection cost UK-wide is that you are pricing three things at once: capacity, authority to act, and evidence afterwards. Cheap plans give you the first. Only the contract tells you whether you are getting the other two, and only the second attack tells you whether you were right. Read the clauses before you compare the prices.

Frequently Asked Questions

How much does DDoS protection cost in the UK per year?

Entry-level proxy plans advertised in the UK market start at modest monthly subscriptions. Mid-market managed services typically land in the low thousands to low tens of thousands annually, depending on protected IP counts and committed clean traffic. Enterprise BGP scrubbing with committed capacity runs higher again, and public sector day rates on G-Cloud listings are quoted per unit per day for incident and consultancy work. Check current pricing directly, as published rates change.

Why is there such a big gap between cheap and enterprise DDoS protection pricing?

Partly technology, mostly people and margin. Cheap plans share automated filtering across thousands of customers with no dedicated engineering time; enterprise contracts commit capacity, a routing relationship and named responders. The resale chain matters too, because each layer between you and the network that owns the scrubbing centres adds margin and removes direct access to the engineers tuning the mitigation.

Does free or bundled protection from a host or CDN count as enough?

For a low-risk brochure site, frequently yes. For anything revenue-bearing, the limits show up fast: little or no tuning, no forensic report, and a provider whose incentive during a large attack may be to null-route your address to protect other tenants. Bundled filtering also rarely covers your authoritative DNS, which is a separate failure point.

What extra charges can appear on an invoice after an attack?

Overage on clean traffic bands is the most common, followed by emergency onboarding fees if you were on an on-demand plan and not yet provisioned. Then look for per-incident support charges beyond included hours, rule change fees above an annual allowance, paid forensic report exports and DNS queries billed by volume.

How do I justify the cost of DDoS protection to finance?

Model revenue or transaction value per hour at peak, add incident staff cost and any liquidated damages you owe customers under uptime clauses, then compare one realistic outage against the annual contract value. Present it as expected loss avoided, not as a doomsday scenario, and cite any national figures you use by source and year so the number survives scrutiny in the room.

Why DDOS Attacks on Government Websites Succeed

Why DDOS Attacks on Government Websites Succeed

A Telegram channel posts a screenshot of a council homepage timing out, the local paper picks it up before lunch, and by mid-afternoon a service owner who has paid for protection for three years is explaining to a director why the site fell over at a traffic volume the provider’s own graph calls unremarkable. That is what most DDoS attacks on government websites look like from the inside. Not a terabit flood. A modest, short, well-aimed burst that found the one hostname nobody had checked.

The public explanations available to a UK public sector team are either definitional (“what is a DDoS attack”) or legal (“it is an offence”) or advisory (“raise your posture”). Useful, none of them answer the question a service owner actually has after an outage: we bought protection, so why did the site go down anyway? The answers are boringly consistent. Leaked origin addresses, unmanaged authoritative DNS, expensive dynamic endpoints, microsite sprawl, and a contract where nobody is contractually obliged to touch a mitigation policy at 02:00.

What an attack on a public sector site actually looks like

Since 2022, the dominant pattern against UK and European public bodies has been pro-Russia hacktivist activity claimed on Telegram: a target list posted in advance or shortly after, availability-only disruption, no attempt at intrusion, and a screenshot of a failed page load as the deliverable. The NCSC published guidance on 25 March 2024 following denial of service attacks against UK political party websites, and has continued to issue advisories about hacktivist groups targeting UK organisations. Treat the pattern as recurring background weather rather than a lightning bolt from the blue.

The screenshot is the payload

This changes the defence economics more than most vendors admit. If the objective is a claim of responsibility rather than sustained damage, the attacker needs five to twenty minutes of visible failure, not five hours. Attacks that fail get quietly dropped from the channel, which is why unsuccessful attempts on high-profile targets tend to vanish from the record; our write-up of hacktivists failing to take the Vatican’s site offline is the shape of an attack that never becomes a story.

Why modest volumes still make national news

A twenty minute outage on a corporate brochure site is a support ticket. The same outage on a council homepage, during a period of political sensitivity, becomes a story about resilience of public services. Journalists do not have your traffic graph. They have a screenshot and a “service unavailable”page.

Confirm it is an attack before anyone briefs communications

Hosting platform failures and DNS provider incidents produce almost identical symptoms to a layer 7 flood, and the first version of the story tends to stick. Establish the difference before you say anything: our walkthrough on telling a DDoS from an outage in ten minutes covers the mechanics. Over-claiming has a cost of its own, because an availability attack that a press office describes as a “cyber attack on resident data”creates a breach narrative you then have to unwind.

What these attacks do not do

A volumetric or application-layer flood does not read your database, exfiltrate records or authenticate to anything. It denies service. Conflating the two in a holding statement invites subject access requests, member questions and an ICO conversation you did not need to have.

Why council and department sites fall over at modest attack volume

Origin IP leakage, the usual culprit

When a proxied site goes down under a few gigabits, the attacker has almost always found the origin. Public sector estates leak origin addresses more than most, because nothing ever gets deleted from a zone file. An MX record pointing at the same server that hosts the site. A staging. Or www2. Host left behind by a supplier in 2019. A consultation subdomain from a 2016 local plan with a live A record on the same /29. Historical DNS datasets and certificate transparency logs make finding these trivial, and no reverse proxy protects an address the attacker can hit directly.

The fix is an audit rather than a product: enumerate every record in every zone you own, resolve each one, and confirm nothing outside the protected set resolves to the origin range. Then firewall the origin to the proxy’s published prefixes and your own management ranges. Most teams find at least one surprise.

Authoritative DNS, the layer outside the contract

Ask a public sector team whether their DNS is protected and a reasonable number will mention Protective DNS. PDNS is a good control and it is not the answer to this question: it protects outbound resolution from staff devices and networks, not the availability of the authoritative name servers citizens query to find you. If those name servers are two boxes in one data centre, or a single provider with a thin anycast footprint, an attack on the DNS layer takes the site off the internet without touching the web tier. The buyer’s questions are in our guide to authoritative DNS DDoS protection.

Endpoints that cost you far more than they cost the attacker

Static pages are cheap to serve and cheap to cache. Site search, planning application lookups, licensing registers, bin collection by postcode, form submissions and large PDFs are not. A few hundred requests per second against a planning search can take down an estate that shrugs off tens of thousands of requests for the homepage. Attack tooling finds these by crawling, and the asymmetry is the point: one request, one uncached database query, one PDF generated on the fly.

Microsite sprawl

This is the structural weakness of the sector. Campaign domains bought for a two-year programme, partnership sites shared with three other authorities, legacy portals from a predecessor body, all with different suppliers, different DNS and inconsistent protection. One unprotected hostname resolving to shared origin infrastructure undoes the protection on everything hosted alongside it. Inventory of hostnames is a security control, not administration.

Check the three surfaces in order: DNS, network, application

DNS

You want an anycast footprint with real geographic spread, a second independent provider (dual-primary, or secondary via zone transfer), and TTL discipline that reflects a trade-off people rarely state out loud: short TTLs give you agility to move records during an incident, while longer TTLs mean resolvers keep serving cached answers if your name servers become unreachable. If you run DNSSEC across two providers, you need a multi-signer arrangement (RFC 8901) or careful key sharing, which is precisely the detail that gets discovered during the incident rather than before it. Our rundown of staying reachable through a DNS provider outage covers the failure modes.

Network and transport

Three honest options. A reverse proxy or CDN suits HTTP services and is the default for most citizen-facing sites, but only works if the origin is genuinely hidden. BGP scrubbing protects whole address ranges including non-HTTP services, and requires that you hold your own IP space and can announce at least a /24, which rules it out for the many authorities sitting on supplier addresses. Hosting-level filtering is the cheapest and most often oversold: ask directly whether “mitigation”at your host means scrubbing or null-routing your IP, because a remotely triggered black hole completes the attacker’s objective on your behalf.

Application layer, and the accessibility decision nobody schedules

Rate shaping per client, per path and per token is where you defend expensive endpoints. Challenge policy is harder, and it is an accessibility decision as much as a security one. A blanket CAPTCHA on a housing benefit or planning objection form excludes the people least able to route around it, and UK public sector sites sit under the Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018, with government guidance now pointing at WCAG 2.2 AA (check the current position before you write a policy). Plan graduated responses: cache aggressively, shape by path, use lightweight non-interactive checks, and keep one tested low-friction route (a phone line, an alternative host, a plain form) that you can publish when you do turn challenges up.

A ten minute check a duty engineer can actually run

  1. Query each authoritative name server directly and non-recursively (dig +norecurse @ns1.example SOA yourdomain.gov.uk). No answer from any of them and the problem is DNS, not the web tier.
  2. Resolve the public hostname from two networks, one of them off your corporate connection.
  3. Curl -sS -o /dev/null -w “%{http_code} %{time_total}\n”against the public hostname, then the same request pinned to the origin with –resolve and the correct Host header. Origin healthy plus public hostname failing points at the edge; both failing points at origin or capacity.
  4. Compare requests per second, bandwidth and 5xx rate on the protection portal’s traffic graph against the same hour last week.
  5. Check the hosting and DNS provider status pages before you declare anything.

What “managed”has to mean in a public sector DDoS contract

The dividing line is simple. If nobody on the supplier side is empowered to change a mitigation policy at 03:00 without waiting for your sign-off, you have bought monitoring, not management. Write it down in the contract: onboarding and rule tuning, always-on baselining, named escalation contacts on both sides, the out-of-hours trigger, and the authority to act.

Trace the resale chain

Plenty of “managed DDoS protection”bought through a hosting partner is resold capacity from a platform two steps removed, and the escalation path runs through two service desks before it reaches an engineer with policy access. Ask whose network, whose scrubbing centres, which regions, and who answers the phone. Get the answer in writing, with names or at least role-based rotas.

Judge providers on evidence, not capacity slides

Aggregate terabits of global capacity tells you little about what is available in the London or Amsterdam scrubbing centre on the day. Press instead on: time-to-mitigate wording in the SLA and what the remedy actually is; whether the SLA covers detection, mitigation or both; the quality of post-attack reports, including vector breakdown, packets and requests per second, sources and mitigation timeline, because that document is what your auditors and any police report will rely on. Ask for a redacted real report before you sign.

Buying routes and the business case

Crown Commercial Service frameworks are the usual route (check which iteration is live when you buy), and multi-supplier estates need one owner of the protection standard across all hostnames rather than each supplier doing its own thing. Build the case from the cost of a day offline: staff time diverted to phones and reception, missed statutory deadlines, delayed payments, member and press handling. Headline capacity numbers make a weak business case; a statutory consultation closing at midnight makes a strong one.

Always-on or on-demand when your service has statutory deadlines

Detection-and-divert means an attack has to be spotted, a route or DNS change made, and traffic drawn into the scrubbing centre before mitigation begins. Against a fifteen minute hacktivist burst, that sequence can complete just as the attack stops. You mitigated nothing and the screenshot already exists. For the citizen-facing hostname, always-on cover is the pragmatic choice even on a tight budget; the wider trade-offs are set out in our comparison of always-on and on-demand protection.

Calendar risk usually settles the argument. The 31 January self-assessment deadline, pre-election periods, GCSE and A-level results days, grant application windows and consultation closing dates are all dates where a thirty minute outage carries legal and reputational weight rather than just annoyance. A hybrid is common and defensible: always-on for the portal and transactional hostnames, on-demand for wider address space and internal ranges where a few minutes of diversion time is tolerable.

Then verify the promise. Run an onboarding test with the provider, agree reporting cadence in peacetime, and ask for evidence of past mitigations at comparable size. A provider who cannot show you a real timeline from a real event is selling you a diagram.

The first hour

Run a fixed sequence rather than improvising: confirm scope (one hostname or the whole estate), check DNS resolution, check reachability to origin and edge, check application behaviour and error mix, then declare and escalate to the named contact. Publish your status update somewhere that does not depend on the failing infrastructure, because a holding page served from the same name servers or the same origin is no use at all. Social channels, a separately hosted status domain and the phone lines all need to be pre-agreed, not improvised while the homepage times out.

Capture evidence as you go: timestamps in UTC, packet or request samples, firewall and proxy logs, mitigation logs and the provider’s attack report. Denial of service is an offence under section 3 of the Computer Misuse Act 1990, with a statutory maximum of up to 10 years, and the National Crime Agency’s position is that using booter or stresser services is itself an offence. Report through Action Fraud, and via your NCSC reporting route as a public sector body; our UK playbook on legal action after an attack sets out what evidence keeps options open. For dated context on how the incident mix shifted, analysis of incidents reported to the ICO and the FCA put denial of service at around a quarter of reported hacking incidents in the first half of 2022, against 4% the year before; the figure is old now, but the direction of travel it showed has not reversed.

Finish with a review that changes something. Re-audit origin exposure, name the owner of every zone and every hostname, close the contract gaps you found at 02:00, and write the triage sequence into the runbook with the escalation numbers on the front page. DDoS attacks on government websites succeed because these small things stay undone between incidents, not because the attackers are formidable. Fix the leaked records, put a second name server provider in place, protect the expensive endpoints, and make sure someone is genuinely on the hook to act before the next target list appears.

Frequently Asked Questions

Why are government websites such frequent DDoS targets?

Because they are symbolic, easy to name and produce disproportionate publicity for a small amount of traffic. Hacktivist groups aligned with geopolitical causes pick targets that make a recognisable screenshot, and a council or department homepage does that better than an anonymous corporate site. The NCSC’s March 2024 guidance, issued after denial of service attacks on UK political party websites, reflects that pattern.

Are DDoS attacks on government websites a data breach?

A denial of service attack disrupts availability; it does not in itself give the attacker access to data. It can still be a reportable security incident, and availability is part of data protection obligations, so take advice on your specific circumstances. Describing an availability attack as a data breach in public creates problems you then have to correct.

Is a DDoS attack illegal in the UK, and who should it be reported to?

Yes. Section 3 of the Computer Misuse Act 1990 covers unauthorised acts that impair the operation of a computer, carrying a statutory maximum of up to 10 years, and the National Crime Agency treats the use of booter and stresser services as an offence in itself. Report to Action Fraud, and public sector bodies should also use their NCSC incident reporting route.

How much does protection for a public sector website cost?

There is no meaningful single figure, and anyone quoting one without asking questions is guessing. Price is driven by the number of protected hostnames, whether authoritative DNS is included, clean traffic volumes, overage rates per Gbps or per million requests, and above all whether managed escalation is real or nominal. Ask for the overage terms and the out-of-hours escalation commitment in writing, because those are where the surprises live.

Does a CDN or reverse proxy alone stop attacks on a council website?

Only if the origin is genuinely hidden and every hostname you own sits behind it. Legacy mail records, staging subdomains and old A records routinely expose the origin address, at which point the proxy is bypassed entirely. Pair the proxy with an origin firewall limited to the provider’s prefixes, and audit your zones for records that resolve straight to the server.

Should a government service use always-on or on-demand mitigation?

For citizen-facing and transactional hostnames, always-on is the safer choice, because hacktivist bursts of five to twenty minutes can finish before a detect-and-divert model has finished rerouting traffic. On-demand remains reasonable for wider address space and internal ranges. If you have statutory deadlines or election periods in your calendar, weight the decision towards always-on for the services that carry them.

Application Layer DDOS Protection: 7 Real Gaps

Application Layer DDOS Protection: 7 Real Gaps

A business that has signed a scrubbing contract, bought a web application firewall (WAF) licence and put a managed service retainer on the renewal calendar tends to file DDoS under “handled”, somewhere between the firewall renewal and the cyber insurance schedule. Then the search page stops answering on a Tuesday afternoon, the origin database sits at 100 per cent CPU, and the edge bandwidth graph never moves. That is the moment most buyers discover that application layer DDoS protection and network-layer capacity are two different purchases, and they only bought one of them properly.

What follows assumes you already have something in place: a reverse proxy, a WAF, a hosting-level filter, maybe all three. The question is not what a Layer 7 attack is. The question is why the thing you are paying for did not stop it.

Why Layer 7 attacks slip past protection you already pay for

An application-layer, or Layer 7, attack sends requests that look legitimate. Valid HTTP, correct headers, often real browsers or real mobile clients driven by malware. The traffic targets whatever endpoint costs the most to serve: a full-text search, a faceted product filter, a login that triggers a password hash, a PDF or invoice render, a GraphQL query that fans out across three services. Volume is beside the point. A few thousand requests per second against a search endpoint can exhaust a connection pool that would shrug off a hundred times that in cached image requests.

So a 40Gbps clean bandwidth commitment tells you nothing about whether 8,000 requests per second to /search gets stopped. Those are different units measuring different failure modes. Network-layer defences count bits; Layer 7 defences have to count requests, weight them by cost, and decide which sessions are real.

Assess three surfaces separately, not as one product

Every estate has three surfaces that fail independently, and contracts routinely cover one or two of them:

  • Network and transport: volumetric floods, reflection and amplification, SYN floods. This is where scrubbing capacity and BGP diversion matter.
  • Application layer: HTTP and HTTPS request floods, slow-read and slow-post attacks, API abuse. This is proxy, challenge and rate-limiting territory.
  • Authoritative DNS: if the name does not resolve, everything above it is decoration.

Write those three headings on a page and map each line of your current contract to one of them. The gaps usually appear within ten minutes.

Read your own telemetry first

The diagnostic signature is unglamorous and reliable. Edge bandwidth flat at normal levels, request counts up sharply on one or two URL paths, origin CPU and database connections climbing, error rates rising on those paths before anywhere else. That pattern points at an application flood, not a pipe-filling attack. If you are still deciding whether you are under attack at all rather than suffering a self-inflicted outage, the triage steps in this guide to telling a DDoS from an ordinary outage will save you arguing about it for an hour.

Gap 1: your origin IP is still reachable, so the proxy is optional

More reverse proxies are bypassed through origin IP leakage than through sheer attack volume. Not close, in practice. An attacker who knows the origin address simply skips the edge entirely and hits the web server directly, and every rule you tuned at the proxy becomes advisory.

The address leaks from predictable places: historical DNS A records captured by passive DNS services long before you onboarded; mail headers from transactional email sent by the same host; TLS certificate transparency logs, which have been mandatory for publicly trusted certificates in Chrome since 2018 and are publicly searchable, so issuing a cert for staging.yourcompany.co.uk or admin.yourcompany.co.uk publishes that hostname to anyone watching; verbose error pages that print internal hostnames; and API endpoints that were pointed direct-to-origin for latency reasons and never moved back.

The fix is dull and effective. Firewall the origin so it accepts inbound traffic only from your provider’s published IP ranges, rotate the origin address after onboarding so historical records are worthless, and then verify it: curl the origin IP directly with the Host header set, from a network outside your own, and confirm you get a refusal rather than a page. Do that before you buy anything else. There is no point evaluating challenge algorithms while the front door has no lock on it.

Gap 2: the WAF is blocking signatures, not behaviour

A WAF ruleset is built to recognise malicious payloads: SQL injection strings, path traversal, known exploit patterns. A Layer 7 flood contains none of those. Each individual request is unremarkable and would be served happily on a quiet Wednesday. Volume in aggregate is the attack, and signature matching has nothing to match.

What actually works sits one layer up in logic: per-session request budgets rather than per-IP counters; JavaScript and cryptographic proof-of-work challenges that cost the client something; and cost-weighted rate limits, which is the control most estates are missing. A request for a cached product image and a request that triggers a full-text search or a PDF render are not equivalent, and a flat limit treats them as if they were. Set it low enough to protect search and you throttle genuine shoppers browsing a catalogue. Set it high enough for browsing and the expensive path stays wide open.

Per-IP limiting also collapses against distribution. A botnet spread across tens of thousands of residential addresses, each sending a handful of requests a minute, never trips a per-IP threshold anywhere. Every source looks like a normal user on a normal broadband line, because it largely is one. The defender’s view of botnet-driven floods is worth reading if you want to understand why source-based blocking degrades as the botnet grows.

One more thing procurement should force into the timeline: the false-positive conversation. Tuning thresholds for your real traffic shape, including scrapers you want to keep and the price-comparison feed you depend on, belongs in onboarding week. Not at 04:00 during an incident, with the commercial director asking why paying customers are seeing a challenge page.

Gap 3: your API and mobile traffic bypass the human-facing defences

Browser challenges, fingerprinting and cookie-based session tracking assume a browser. Your mobile app does not have one. Neither does your partner’s integration, your payment webhook receiver, or the B2B client pulling stock levels every thirty seconds. Turn on aggressive challenges and you break them; exempt them and you have created an unprotected lane straight to the most expensive endpoints you own.

Authenticated APIs are the attractive target precisely because they skip caching and touch the database on every call. Checkout APIs, GraphQL endpoints that allow deeply nested queries, and account-history calls all cost real compute per request. Sensible controls here are different in kind: rate limits keyed to API tokens or account identity rather than IP; explicit request cost budgets where a heavy query consumes more of the allowance than a light one; schema validation and query depth limits on GraphQL; and mutual TLS or signed requests for partner traffic so you can distinguish it cleanly.

Check subdomain coverage while you are there. One forgotten host, a legacy api-v1 or a marketing microsite on the same origin, undoes the proxy configuration for everything behind it.

Gap 4: authoritative DNS sits outside the contract

Layer 7 protection and DNS protection are frequently sold as one story and scoped as two line items. Read the schedule. Authoritative DNS is the layer most often left unmanaged, usually because it came free with the registrar years ago and nobody revisited it.

Ask who answers queries for your zone when your primary DNS provider is itself the target, how many anycast points of presence serve UK resolvers, and whether secondary DNS with a second independent provider is included or chargeable. Then look at your TTL values. A 3,600 second TTL means up to an hour before a failover to a backup provider takes effect across resolvers; 60 to 300 seconds, set in advance and left there, gives you near-immediate redirection when you need it. Lowering TTLs during an incident does not help, because the old values are already cached.

Our buyer’s guide to authoritative DNS protection goes through anycast footprints and secondary configurations in more detail, and the record of what happens when a DNS provider itself goes down explains why single-provider dependency keeps producing the same headline.

Gap 5: nobody is contractually obliged to act during the attack

The dividing line is simple. A self-service plan gives you a dashboard, a rate-limiting engine and rules you write and troubleshoot yourself. A managed service means a named engineer on the provider’s side is empowered to change policy at 03:00 without waiting for your sign-off. If that authority is not written down, you have bought monitoring, not management.

Deployment model matters more at Layer 7 than most buyers expect. On-demand mitigation, where traffic is diverted via BGP once an attack is detected, is a defensible trade-off for volumetric floods; detection plus diversion plus route convergence is tolerable when the alternative is paying for always-on capacity you rarely use. It is a poor fit for application attacks. A 5,000 requests-per-second flood against a search endpoint can drain your database connection pool well before diversion completes and traffic starts flowing through the scrubbing path. By the time mitigation is live, you have already taken the outage. The comparison of always-on and on-demand models sets out where each one genuinely earns its cost.

In procurement, run the escalation test rather than reading the SLA summary. Ask who answers the phone at 03:00 on a Sunday, whether that person is in the UK or following the sun, what they are authorised to change unilaterally, and how a customer raises severity. Then look at the wording: time to mitigate versus time to first human response (they are not the same metric), what counts as an attack start for the purposes of the clock, and the exclusions. Many time-to-mitigate commitments apply only to network-layer events and quietly exclude application-layer mitigation altogether.

Gap 6: you are buying a resale chain, not a network

A large number of UK providers resell another vendor’s proxy or scrubbing platform under their own brand. That is not automatically a problem; the underlying platforms are often excellent, and the reseller may add genuine tuning value. The problem is the extra escalation hop. During an incident, the distance between you and the engineer who can actually push a rule change is measured in handoffs, and each one costs minutes you do not have.

Questions that tend to get honest answers: which points of presence serve UK traffic and where do they sit; what is the underlying platform; who holds the capacity contract; can your engineers speak directly to the platform’s operations team during a severity-one event, or does everything route through account management. If the answers are vague, the escalation path is probably vaguer.

Gap 7: no post-attack evidence, so nothing improves

Judge a provider on the quality of a redacted post-incident report, not on a headline Tbps figure. Anyone can print a capacity number. A report that names the endpoint under attack, the rule that fired, the time from detection to mitigation and the residual request rate still reaching your origin tells you whether there was an engineer watching.

A usable attack report contains request rates broken down by endpoint and method, source distribution by network and geography, the vector breakdown, which mitigation action was applied and when, and what got through. That evidence has value well beyond engineering. Cyber insurers ask for it, boards ask for it, and any legal route following an attack depends on contemporaneous logs rather than recollection. The UK’s National Cyber Security Centre sets out the same expectation in its denial of service guidance collection: know what normal looks like, and capture what happened.

On the business case, argue from downtime cost per hour, not the headline monthly fee. UK managed pricing is typically assembled from four components: clean bandwidth commitment, the number of protected assets or domains, request volume tiers, and the level of incident support. Those variables move independently, so two quotes with similar monthly figures can differ enormously in what they cover at Layer 7. Work out revenue and recovery cost for a four-hour outage during your busiest trading window, and the comparison becomes straightforward.

The pre-purchase checklist

  1. Curl your origin IP directly from an external network with the Host header set. Do you get a page?
  2. Ask for the endpoint-level rate-limiting configuration, weighted by request cost, that the provider proposes for your top five most expensive paths.
  3. Confirm in writing which mobile and API traffic is covered, and how, given those clients cannot solve browser challenges.
  4. Check whether authoritative DNS is inside the managed scope, and whether secondary DNS is included.
  5. Establish who changes rules at 03:00, their authority level, and the time-to-mitigate definition that applies to Layer 7 specifically.
  6. Ask what the underlying platform is and how many escalation hops sit between you and it.
  7. Request a redacted Layer 7 post-incident report as a sample deliverable before you sign.

If you are still mapping requirements before shortlisting, our overview of DDoS attack solutions and how they differ covers the architectural options in more detail. Effective application layer DDoS protection is rarely a product decision. It is a coverage decision, an authority decision and an evidence decision, and the estate that gets those three right will survive a request flood that flattens a better-funded competitor.

Frequently Asked Questions

What is the difference between application layer DDoS protection and network layer scrubbing?

Network-layer scrubbing removes volumetric and protocol attacks measured in bits and packets per second, usually by diverting traffic through a scrubbing centre. Application layer DDoS protection inspects HTTP and HTTPS requests, counts them per endpoint and per session, and decides which clients are genuine. A contract covering one does not necessarily cover the other, and the SLA clocks are often defined separately.

Can a WAF alone stop a Layer 7 DDoS attack?

Rarely. A WAF is built to recognise malicious payloads, and a request flood consists of well-formed, individually legitimate requests. Stopping it needs behavioural controls: session-level request budgets, challenges, bot scoring and cost-weighted rate limits on your most expensive endpoints. A WAF with those features enabled and tuned can do the job; a signature-only ruleset will not.

How do attackers find my origin IP and get past the proxy?

Through historical DNS records held by passive DNS services, mail headers from transactional email sent by the same host, certificate transparency logs that publish staging and admin hostnames, verbose error pages, and API endpoints pointed straight at the origin. The countermeasure is to restrict inbound traffic at the origin firewall to your provider’s published IP ranges and rotate the origin address after onboarding.

Does application layer DDoS protection cover APIs and mobile app traffic?

Only if it was scoped to. Browser challenges and JavaScript fingerprinting do nothing for clients that cannot execute JavaScript, so API and mobile paths need token-based limits, request cost budgets, schema and query depth validation, and signed or mutually authenticated requests. Ask the provider specifically how non-browser clients are handled, and check every subdomain is behind the edge.

What does managed application layer DDoS protection typically cost in the UK?

Pricing is usually assembled from clean bandwidth commitment, the number of protected domains or assets, request volume tiers and the incident support level, so published headline figures are hard to compare and change frequently; check current quotes directly. The more useful exercise is calculating your cost of downtime per hour during peak trading and testing the annual protection spend against that. Get each quote broken down by those four components so you are comparing the same coverage.



Authoritative DNS DDOS Protection: A Buyer's Guide

Authoritative DNS DDOS Protection: A Buyer’s Guide

A UK retailer can have its web front end behind a reverse proxy, a signed scrubbing contract with a named escalation matrix, a web application firewall tuned over months, and still go completely dark for two hours because two nameservers in the same rack stopped answering queries. The firewall logs will look healthy. The scrubbing dashboard will show nothing. Authoritative DNS DDoS protection is the part of the estate most buyers assume is included in something else, and it almost never is.

Authoritative DNS is the service that answers “what is the address for this name”for zones you control. Not the resolver your laptop talks to. The servers listed in your NS records, the ones every recursive resolver on the internet eventually has to reach. Take them offline and the name stops existing, whatever else is still running.

Why authoritative DNS fails differently from your website

What actually stops when the nameservers stop

A website outage is visible and bounded. A delegation outage is broader than most incident plans assume: inbound mail stops because sending servers cannot resolve your MX records; API clients fail on lookup rather than on connection; VPN concentrators become unreachable to remote staff resolving the endpoint hostname; SaaS single sign-on breaks in both directions as identity providers and service providers try to reach each other by name; and payment provider callbacks to your webhook endpoints fail silently, which usually surfaces later as a reconciliation problem rather than an alert.

Support desks feel it before monitoring does. Staff report “the internet is down”, which sends the first responders looking at connectivity instead of at the zone.

TTLs decide whether you notice in an hour or in a minute

Resolver caching is the only shock absorber you get. A record with a time-to-live (TTL) of 3600 seconds sits in resolver caches for up to an hour, so an authoritative outage may be invisible to a large share of users while you recover. Set the same record to 300 seconds because you wanted fast failover, and caches expire five minutes into the attack. Resolvers then come back for fresh answers, find nobody home, and the outage is immediate and total.

That is a genuine trade-off, not a best practice with one right answer. Long TTLs buy grace and take away your ability to move traffic quickly under pressure. Short TTLs do the reverse. Most estates end up with a split: 3600 seconds on stable A and MX records, 300 seconds only on the handful of records involved in a planned failover, with the low values set in advance of a change window rather than left permanently low.

The DNS version of origin IP leakage

Origin leakage is a familiar failure: the web estate hides behind a proxy, but the origin address is discoverable, so the attacker skips the proxy. The same pattern repeats one layer down, and it is more common than the web version. The front end is proxied and filtered. The NS records still point at ns1 and ns2 on the same unfiltered transit, frequently in the same rack, sometimes in the same /24. An attacker reading the delegation does not need to find the origin. It is published.

This is the single most useful check in the whole exercise, and it takes thirty seconds. Query your NS records, resolve the nameserver hostnames, and ask whether those addresses sit behind the same protection you bought for the website. If they do not, the scrubbing contract is protecting the part of the estate nobody needs to attack. Our earlier piece on the gaps that commonly remain open on DNS infrastructure covers the operational side of that audit.

The attack patterns authoritative DNS protection has to survive

Direct query floods over UDP and TCP

The straightforward case is volume: spoofed UDP queries crafted to resemble ordinary recursive traffic, aimed at your nameserver addresses. Because the source addresses are forged and the queries are well formed, signature matching helps less than capacity and per-source behavioural limits do.

TCP query floods get overlooked. DNS over TCP is a requirement, not an edge case, as set out in RFC 7766, and resolvers fall back to TCP routinely for large responses. An attacker opening connections and abandoning them exhausts connection state long before it exhausts bandwidth. Ask any prospective provider how it handles TCP query state separately from UDP volume. The vague answers are informative.

Random subdomain floods break the usual blocking reflex

Water torture attacks query nonexistent labels under your zone at high rates: a random string, then your domain. The queries arrive at your authoritative servers through legitimate recursive resolvers, because that is how the internet works. Your servers do the NXDOMAIN work, and if DNSSEC is in play, the negative answers cost more to produce.

The instinct is to block the sources. Do that and you have blocked BT’s resolvers, Sky’s resolvers and whichever public resolver a third of your customers use. The meaningful procurement questions are narrower: how does the platform absorb NXDOMAIN load, can it rate limit on label patterns rather than source addresses, and will it show you the top queried labels after the event so you can prove what happened? Response rate limiting in BIND, NSD and Knot helps on self-hosted servers, but it is a blunt instrument against traffic arriving via shared resolvers.

Your zone as somebody else’s weapon

The other direction matters for reputation and for your own capacity. Large ANY or TXT responses make your nameservers a useful reflector: an attacker spoofs a victim’s address, sends small queries, and your servers deliver the amplification. RFC 8482, published in January 2019, defines minimal-sized responses to ANY queries precisely to close that down. Check whether your provider implements it, and check what your TXT records have accumulated over the years. Old SPF fragments and abandoned verification strings add up.

Delegation failures that look like an attack

Some DNS outages are control failures wearing an attack costume. Hijacked NS records after a registrar account compromise. A domain that expired because the renewal invoice went to a departed employee. A DNSSEC key rollover where the DS record at the registrar no longer matches the signing key at the provider, which makes validating resolvers return SERVFAIL for a zone that is publishing perfectly good answers.

For generic top-level domains, registrar-level EPP statuses such as clientUpdateProhibited and clientTransferProhibited are the cheapest control you can apply. Nominet runs.uk under its own rules, so ask your registrar what lock equivalents it supports rather than assuming the gTLD status codes apply. And remember where the DS record lives: at the registrar, in the parent zone, not with your DNS provider. That split is what turns provider migrations into multi-day incidents.

Deployment models compared: where each one fits

Managed anycast DNS as a hosted service

You hand over the zone and inherit a global anycast footprint immediately. Capacity gain is the fastest available. What you give up is granularity: you rarely choose the rate-limiting thresholds, you get the logging the platform offers, and your ability to change records during an incident depends entirely on that provider’s portal and application programming interface (API) staying up.

Self-hosted nameservers behind BGP scrubbing or hosting-level filtering

Full control of software versions, response policy, rate limiting and logs. The catch is timing. On-demand mitigation fits nameservers badly, and diversion time is only half of the problem. If your records carry 300 second TTLs, resolvers come back for answers during the exact window you are still moving routes and settling the scrubbing path. Always-on is usually the honest answer for the DNS surface, whatever you decide for web traffic. The wider trade-off is covered in our comparison of always-on and on-demand mitigation models, and the measurement problem in our piece on what time-to-mitigate figures in an SLA really mean.

Dual-provider authoritative DNS with a hidden primary

The stronger architecture, and the one most organisations with real availability obligations end up at: a primary server you control, not published in the NS set, transferring the zone by AXFR and IXFR to two independent public secondary networks. Both providers answer queries. Neither one holds your only copy of the zone or your only editing path.

It costs discipline. Transfer monitoring, TSIG key management, DNSSEC handled either by signing at the hidden primary or by keeping both providers aligned, and a habit of checking that both NS sets serve the same serial. Compare that with the common single-provider setup where one portal outage removes both resolution and your ability to change anything. October 2016 is still the reference case: the Mirai-driven DNS provider outage attack pattern that hit Dyn took large numbers of otherwise healthy sites off the internet, and Dyn’s own incident write-up, published later that month, described tens of millions of discrete IP addresses associated with the botnet traffic. The sites were fine. The delegation was not.

Matching the model to the estate

Single-site ecommerce with one zone and a small team: managed anycast, always-on, with the registrar locked. Multi-tenant hosting or a data centre carrying customer zones: dual-provider or self-hosted with always-on filtering, because your customers’ NS records are your reputation. Healthcare, government services and anyone with regulatory availability expectations: dual-provider with a hidden primary, and a tested delegation rollback, because “our supplier was attacked”is not a defence that survives an availability review.

What ‘managed’ actually buys on the DNS surface

Pin down four things in writing, because sales decks cover none of them properly: onboarding and zone import, including who validates parity after the import; tuning of rate limits and query filters, and whether you can request changes; the escalation path during an attack, with named contacts and out-of-hours cover; and who is permitted to change records while an attack is in progress.

That last point is where recovery actually stalls. If the portal is degraded, is there an API? If there is an API, what is its rate limit when you need to push forty record changes at 02:00? Will the provider act on your behalf without sign-off, and is that authority written down? If nobody on their side can change policy or a record out of hours without your director’s approval, you have bought monitoring, not management.

Reporting quality is the second tell. A one-line “we mitigated 40Gbps for you”is useless on DNS. Ask to see a sample report before you sign, and look for queries per second over time, NXDOMAIN ratio against baseline, top queried labels, query type breakdown and a source resolver breakdown. If the provider cannot produce that shape of report, it cannot diagnose a water torture attack either.

Then trace the network. Provider diversity is frequently illusory. Two separately branded DNS services can resolve back to the same anycast footprint or the same upstream scrubbing partner, which means your carefully designed redundancy shares a single failure domain. Ask which network operates the points of presence, which autonomous system numbers announce the anycast prefixes, and whether any part of the mitigation is subcontracted. Resale is not a scandal, but you should know you are buying it. Our guide to choosing managed DDoS protection in the UK goes further into how those resale chains are structured and priced. Law firms weighing these same resale chains will find sector-specific buyer advice in our piece on keeping client portals, DNS and websites online beyond phishing.

Evidence to demand before you sign

  • Capacity in queries per second, not just Tbps. Terabits describe a pipe; nameservers fall over on query rate and on NXDOMAIN work. Ask for the qps figure, how it was measured, and whether that headroom is per customer or shared across the platform.
  • An SLA that covers query response availability. Portal uptime is not resolution uptime. Read what the credit pays against, then compare it with your own downtime cost per hour. Most credits are rounding errors by that measure.
  • Time-to-mitigate stated for the DNS layer specifically, with always-on versus on-demand made explicit and your TTL values factored into the conversation.
  • Control checks. Registrar lock status, who holds the DNSSEC signing keys, whether the provider supports a hidden primary and inbound zone transfers, and a documented, tested rollback of the delegation.

Costs, business case and a safe cutover

UK pricing on this surface tends to follow one of three shapes: per zone with a query allowance, per million queries with overage, or bundled into an existing scrubbing or proxy contract at a marginal rate. Overage is where the surprises live, because an attack generates query volume you did not plan for. Ask whether attack traffic is billable. Pricing on managed DNS moves, so confirm current figures directly with the provider rather than working from anything published last year.

Build the case from your own numbers, not from headline attack sizes. Lost orders per hour from your own order data, staff hours idled, SLA credits you would owe your own customers, and the support volume an outage generates for weeks afterwards. Set that against annual protection spend plus plausible overage. The first-hour ecommerce response playbook is a useful companion for working out what an hour actually costs you.

The cutover itself breaks more zones than attacks do. A sequence that works: run a full record parity check between old and new providers and diff the output; lower TTLs on the records you are moving several days ahead, and wait out the old TTL; add the second NS set at the registrar while keeping the first; watch resolution from several independent networks and public resolvers, not from your office; then restore longer TTLs once things are stable. If DNSSEC is signed, plan the DS record change separately and do not remove the old keys until validating resolvers have picked up the new chain.

Afterwards, keep three checks running: external resolution from multiple networks, zone transfer health and serial agreement between providers, and DS record consistency against what your provider is publishing. Authoritative DNS DDoS protection is bought once and verified continuously, and the verification is what separates an estate that stays reachable from one that discovers its delegation problem during the incident. More background on attack infrastructure and defender priorities is collected across the DDoSInfo archive.

Frequently Asked Questions

What is the difference between protecting authoritative DNS and protecting recursive DNS?

Authoritative DNS answers queries about zones you own, so an outage makes your domain unreachable to everyone on the internet. Recursive DNS is the resolver service your users and servers query, so an outage there affects your own people’s ability to reach anything. Both need protection, but the blast radius and the buying decision are different: authoritative is a public-facing availability problem, recursive is an internal service problem.

Does a reverse proxy or CDN already give me authoritative DNS DDoS protection?

Only if you have moved your authoritative nameservers to that provider as well. A proxy in front of your web origin does nothing for queries aimed at the addresses in your NS records. Check the delegation: resolve your nameserver hostnames and confirm those addresses sit behind filtering. If they point at self-hosted servers on ordinary transit, the protected front end is not the exposed surface.

Is always-on mitigation necessary for nameservers, or will on-demand do?

For authoritative DNS, always-on is usually the right answer. On-demand requires detection plus diversion time, and if your records carry short TTLs for failover, resolver caches expire during that window and the outage becomes immediate rather than gradual. On-demand can be defensible for a web estate with long cache lifetimes; it fits the delegation badly.

Do two DNS providers actually make me safer, and how do I check they are independent?

Two providers remove single-provider dependency for both resolution and record editing, which is the failure the October 2016 Dyn incident illustrated. They only help if they are genuinely separate. Ask each provider which network operates its anycast points of presence, which autonomous systems announce the prefixes, and whether mitigation is subcontracted to a third party; shared upstreams mean shared failure.

How does DNSSEC affect DDoS protection and failover between providers?

DNSSEC adds signing work and larger responses, which raises the cost of NXDOMAIN floods against your zone, and it makes key handling part of your availability design. The DS record lives at the registrar in the parent zone, not at your DNS provider, so a mismatch between the published DS and the active signing key causes validating resolvers to fail the zone outright. Multi-provider DNSSEC works, most reliably with signing done at a hidden primary you control, but treat every key rollover as a change with its own rollback plan.

Terabit DDOS Attack: What It Means for Your Site

Terabit DDOS Attack: What It Means for Your Site

Every few months a mitigation provider publishes a traffic graph with a new peak on it, the trade press repeats the figure, and somewhere in the UK a head of infrastructure forwards the link to their team with one line attached: are we covered for this? A terabit DDoS attack makes an excellent headline and a poor planning document. For a mid-market ecommerce or SaaS platform running a few hundred Mbps of normal traffic, with occasional spikes to a couple of Gbps, the record figures change almost nothing about what should be built. The things that actually take those sites offline are smaller, quieter and mostly unrelated to volume records.

What follows is the reasoning behind that claim, and then the specific evidence to demand from any provider whose sales deck leads with a terabit number.

What a terabit DDoS attack actually is, and where the traffic lands

A distributed denial-of-service (DDoS) attack at terabit scale means aggregate inbound traffic measured in terabits per second (Tbps), typically a flood of User Datagram Protocol (UDP) packets sent from tens of thousands of compromised hosts towards a target address or prefix. Three numbers describe such an event and they are not interchangeable: peak bandwidth in Tbps, packet rate in packets per second (pps), and total bytes delivered over the life of the attack.

Peak Tbps sells press releases. Packet rate is what kills stateful devices, because a firewall or load balancer has a pps ceiling long before it has a bits-per-second ceiling. Total bytes delivered is the honest measure of how much traffic a network really absorbed, and it is usually the number missing from the headline. Cloudflare’s disclosure of a 7.3 Tbps flood in mid-2025 put it at roughly 37.4 terabytes delivered in about 45 seconds. Short, brutal, over before most on-call rotas have finished reading the first alert.

Reading the record numbers properly

Publicly reported peaks have climbed quickly. Cloudflare reported 7.3 Tbps in mid-2025, then 11.5 Tbps later in the year, and by late 2025 its own quarterly DDoS threat reports had pushed the disclosed peak past 20 Tbps, with the trend line heading towards 30 Tbps. Several of the largest lasted under a minute. Duration matters more than most buyers assume, and it is the detail that reframes the whole on-demand mitigation question later in this piece.

The other detail worth holding onto: these measurements are taken at a global anycast edge spread across hundreds of points of presence. Anycast means the same IP address is announced from many locations at once, so a botnet firing at one address is automatically split across dozens of scrubbing points by internet routing. No single facility receives the full volume, and no single customer origin ever sees it. Treating a 30 Tbps figure as “the flood that would arrive at my rack”is a category error, not a conservative assumption.

Where the volume is absorbed

Terabit-scale traffic is eaten by upstream transit and peering capacity, by anycast distribution across scrubbing locations, and by filtering hardware sitting in those locations. None of that sits in your cabinet. Your firewall, your load balancer and your origin uplink are downstream of every decision that determines whether a large flood is survivable, which is why capacity is the one part of the problem you genuinely cannot fix yourself.

The sources behind the biggest floods are consistent: large botnets of Internet of Things (IoT) devices, compromised consumer routers and customer premises equipment, plus reflection and amplification through exposed UDP services. The economics are well documented in the booter and stresser market, where rented capacity is sold by the minute; the practical defender response to botnets available for hire has little to do with matching their peak output.

Why volume records rarely explain why a UK site went down

Run the arithmetic on your own estate and the record numbers lose their grip. A 10 Gbps origin uplink is saturated by roughly 0.03 percent of a 30 Tbps event. A 1 Gbps uplink goes under at about 0.003 percent. Anything above a few Gbps pointed at a single origin is already a complete outage, so the difference between a 7 Tbps record and a 30 Tbps record is academic from where you sit. Your real ceiling is three to five orders of magnitude below the league table.

So what does take these sites down?

Origin IP leakage, the most common bypass

More proxy bypass is caused by leaked origin addresses than by raw volume. The usual culprits are boring and repeatable: outbound mail headers that stamp the sending host’s real address on every transactional email; historical DNS records sitting in passive DNS databases years after a migration; certificate transparency logs listing staging, dev and admin subdomains that resolve straight to the origin; monitoring probes and payment callbacks configured against the origin IP because the proxy interfered with a health check once in 2021.

A terabit capability is irrelevant when 2 Gbps arrives at an address that was never supposed to be public. The fix is unglamorous: firewall the origin so it accepts traffic only from your provider’s published ranges or a private cross-connect, rotate the origin IP after onboarding, and send mail through a separate relay.

Application-layer floods that never register in Tbps

Layer 7 attacks are measured in requests per second, and the good ones are surgical. Consider 120,000 requests per second against an uncached product search endpoint on a mid-sized ecommerce stack. Database connections exhaust, queue depth climbs, the site returns 502s, and the whole thing registers well under 1 Gbps. Invisible in any terabit ranking. Cart, login, coupon validation and faceted search are the endpoints attackers reach for, because each request costs the target far more than it costs the botnet. The first sixty minutes of an ecommerce DDoS incident almost always involve this kind of traffic rather than a wall of UDP.

Authoritative DNS, the layer nobody owns

Authoritative Domain Name System (DNS) service is the surface most often left with whatever the registrar or hosting control panel supplied at setup. It is also the single point where a modest query flood removes every service at once, including the well-protected ones, because a client that cannot resolve your name never reaches your carefully scrubbed web tier. Three questions settle it: who operates your authoritative DNS, across how many independent anycast networks, and is that service inside the same mitigation contract as your web traffic? If the answer to the third is no, sound DNS DDoS attack protection is the cheapest resilience upgrade available to most UK buyers, and the specific gaps to close at the DNS layer are worth reviewing before renewal.

Deployment models that can absorb terabit-scale volume, and what each costs you

Four models dominate the UK market, and each buys capacity in a different currency.

  • Reverse proxy on anycast. Very large aggregate capacity, fast HTTP and HTTPS mitigation, straightforward onboarding by DNS change. Protects only what resolves through it, and collapses as a control the moment the origin is directly reachable.
  • Border Gateway Protocol (BGP) scrubbing and diversion. Protects whole prefixes and non-HTTP services such as SMTP, game protocols and APIs on odd ports. Needs a /24 or larger that you control, plus a return path by Generic Routing Encapsulation (GRE) tunnel or cross-connect, and route propagation adds minutes you must plan around.
  • Hosting-level and data centre filtering. Convenient, frequently bundled, sometimes genuinely capable. Capacity is shared across every tenant and the ceiling is whatever the facility’s upstream transit contract allows, which is rarely disclosed in the order form.
  • On-premise appliances. Useful for application-layer inspection and internal east-west protection. Useless against volumetric floods, because the flood arrives on a pipe the appliance cannot widen.

Always-on versus on-demand against a 40-second flood

Here is where the record-breaking figures earn their place in the analysis. A flood lasting 35 to 45 seconds exposes the on-demand model’s central weakness. Detection has to fire, a human has to authorise diversion, a BGP announcement has to go out, and routes have to propagate across the internet. That sequence can consume more time than the attack lasts, producing the graph nobody wants in the post-incident pack: mitigation starting just as the traffic stops.

The recommendation is blunt. Always-on for public web surfaces, and on-demand reserved for cases where diversion genuinely costs you in latency or money, such as a latency-sensitive trading path or a rarely attacked internal prefix. The trade-offs between always-on and on-demand protection are real, but sub-minute bursts have tilted the balance decisively for anything customer-facing.

How to read a provider’s terabit capacity claim without a lab

Capacity claims come in three flavours and only one of them affects your uptime.

  1. Aggregate network capacity. The sum of everything, everywhere. Marketing number.
  2. Per-point-of-presence capacity. More honest, still not yours.
  3. Capacity in the region serving your users. For a UK customer base, that means London and, ideally, a second European site for failover. This is the only figure that decides whether your Tuesday morning stays quiet.

Then trace the resale chain. A meaningful share of UK managed DDoS protection is resold, sometimes twice, and the capacity in the deck may belong to a network the seller has no operational control over. Ask directly: whose autonomous system number announces my prefix during mitigation, whose scrubbing centres filter it, and which company employs the engineers who join the bridge at 03:00? A reseller can be a perfectly good buy if the answers are clear and the escalation path is named. It is the vagueness that should worry you.

Evidence to request instead of a slide

  • A redacted attack report from the last quarter showing vector breakdown, peak figures, detection timestamp and mitigation timestamp;
  • The service level agreement (SLA) definition of time to mitigate, with confirmation that the clock starts at detection rather than at ticket acknowledgement, a distinction covered in more depth in this examination of what SLA mitigation seconds actually mean;
  • A named escalation matrix with roles and out-of-hours numbers, not a shared inbox;
  • Change-freeze behaviour during peak trading, including who can push a rule change on Black Friday and who signs it off.

What “managed”has to mean in writing

Managed is the most elastic word in this market. Insist on written detail for onboarding and initial rule tuning, always-on traffic baselining, who is authorised to issue a mitigation order without customer sign-off, how a false positive gets rolled back mid-flash-sale, and what a post-incident report contains and when it arrives. If nobody on the provider’s side can change policy at 03:00 without waking your CTO, you have bought monitoring, not management. The buyer’s guide to managed DDoS protection in the UK goes further on how these contracts are typically priced and scoped.

Sizing your own exposure and building the business case

Work out your real ceiling in this order: origin port speed, upstream transit commit (the committed rate, not the burst), load balancer packet-per-second throughput, then authoritative DNS query capacity. Most teams discover the binding constraint is not the number they assumed. A 10 Gbps port fronted by an appliance that tips over at a few hundred thousand packets per second has a pps ceiling, not a bandwidth ceiling.

UK pricing tends to be built from committed clean bandwidth, per-prefix or per-domain charges, overage on attack traffic, and a cap on mitigation events per year. Read the overage clause carefully, because a single large event billed on attack traffic can dwarf the annual subscription.

Build the case from downtime cost, not from last quarter’s record. Do the division on your own trading figures: a shop turning over roughly £40,000 a day is losing, on that arithmetic alone, in the region of £1,700 of gross revenue per hour offline, before chargebacks, support volume and paid advertising spend burned on landing pages nobody can reach. That figure justifies the spend. A 30 Tbps headline does not.

One scenario to rehearse before it arrives: the ransom demand, usually a short demonstration burst followed by a deadline and a wallet address. Escalate to your provider and report it to Action Fraud, the UK’s national fraud and cybercrime reporting centre. Never pay. Payment marks the target as willing and invites the next demand, a pattern seen repeatedly in extortion cases involving DDoS blackmail prosecutions.

A short exposure review to run this month

  1. Leak hunt: check certificate transparency logs, passive DNS history and outbound mail headers for your origin address, then lock the origin to provider ranges only;
  2. DNS resilience check: confirm who runs authoritative DNS, how many anycast networks carry it, and whether it sits inside your mitigation contract;
  3. Failover test: divert and return during a maintenance window, and time it properly against the SLA;
  4. Escalation contact: one named person at the provider, one deputy, tested out of hours, documented where the on-call engineer can find it at 03:00.

The UK’s National Cyber Security Centre publishes practical guidance on denial of service preparation that pairs well with that checklist, particularly on understanding your own service dependencies before an incident rather than during one.

Frequently Asked Questions

How big is the largest terabit DDoS attack recorded so far?

Publicly reported peaks moved from 7.3 Tbps, disclosed by Cloudflare in mid-2025, through 11.5 Tbps, to figures above 20 Tbps in Cloudflare’s own reporting by late 2025. All of these are vendor-reported measurements taken at a global anycast edge, and the record moves often, so treat any figure as accurate at the time of writing rather than permanent.

Would a terabit DDoS attack take down a site that already has protection?

If the site is behind a large anycast network and the origin is not directly reachable, a terabit flood is absorbed across hundreds of locations and no single facility sees the full volume. The realistic failure modes are different: a leaked origin IP, an application-layer flood aimed at expensive endpoints, or an unprotected authoritative DNS zone. Fix those three before worrying about the peak figure.

How much bandwidth does it take to knock a typical business website offline?

Far less than the headlines suggest. A 1 Gbps origin uplink saturates at roughly 0.003 percent of a 30 Tbps event, and a 10 Gbps uplink at roughly 0.03 percent, so anything above a few Gbps arriving at a single unprotected origin is already a full outage. Stateful devices often fail earlier still, on packet rate rather than bandwidth.

Can on-demand DDoS protection stop a 40-second terabit flood?

Usually not in time. Detection, human authorisation, BGP announcement and route propagation can together take longer than a 35 to 45 second burst, so mitigation engages as the traffic ends. Always-on filtering is the sensible default for public web surfaces, with on-demand kept for prefixes where diversion carries a genuine latency or cost penalty.

What evidence should I ask for when a provider claims terabit-scale scrubbing capacity?

Ask for capacity in the region serving your users rather than the global aggregate, a redacted attack report from the last quarter with detection and mitigation timestamps, and an SLA that starts the time-to-mitigate clock at detection. Then trace the resale chain to the party that owns the scrubbing points and employs the on-call engineers. If a claimed terabit DDoS attack capability cannot be attached to a named network, named facilities and a named escalation path, treat the number as somebody else’s. If that SLA is ever tested in anger, see our UK playbook for legal action after a DDoS attack for evidence and claims steps.

DNS Provider Outage Attack: How to Stay Reachable

DNS Provider Outage Attack: How to Stay Reachable

A DNS provider outage attack is the one denial-of-service scenario your own capacity planning cannot touch: the origin is healthy, the load balancers are bored, the CDN edge is warm, and the domain is still dark because somebody else’s authoritative nameservers are drowning. Nothing in your rack is broken. There is no traffic to divert, no scrubbing centre to call in, no BGP announcement to shift, because the network under pressure belongs to a third party you pay a few hundred pounds a year and have probably never spoken to on the phone.

Most coverage of this topic retells the October 2016 Dyn attack and stops. That is a story, not a control. What follows treats authoritative DNS as what it actually is: an architecture and procurement decision that gets made once, usually by accident, and then sits unexamined until the day it fails.

What a DNS provider outage attack actually breaks

Authoritative DNS is the set of servers that hold the definitive answers for your zone. Recursive resolvers, run by ISPs, mobile networks, corporate estates and public services such as Google, Cloudflare and Quad9, ask those servers on behalf of users and cache the reply. Attack the authoritative layer and you have not touched a single one of your machines. You have simply removed the directory that tells the internet where they are.

Three patterns fail in different ways, and the distinction matters because the remedies differ:

  • Straight query floods exhaust anycast capacity at the provider’s points of presence. Packets per second, not clever queries. The provider’s headroom is the only thing standing between you and a timeout.
  • Pseudo-random subdomain floods, often called water torture, query names like a8f3k2.yourdomain.co.uk in volume. Every one is a genuine miss. No resolver cache can absorb them, so the load is funnelled through recursive resolvers onto the authoritative servers, and the resolvers themselves fill with useless negative entries. The attack borrows the legitimate resolver population as a delivery mechanism.
  • Reflection and amplification using spoofed source addresses can point the reflected volume at the authoritative infrastructure, or at you, depending on which end of the spoof you are on.

Then comes the part that lengthens every DNS incident: retries. When answers time out, resolvers do not politely wait. They re-query, try the next nameserver in the set, and try again, which means legitimate traffic multiplies exactly when the provider is trying to recover. The IETF eventually wrote guidance about it. RFC 9520, published in December 2023, tells resolvers to cache resolution failures for at least one second and, as guidance, no longer than five minutes, specifically to damp down retry storms against servers that are already struggling. If those retry storms start to resemble an assault, a signal-by-signal triage can distinguish an outage from a genuine DDoS within minutes.

Everything keyed to that zone goes at once. Not just the website: API endpoints partners depend on, MX records and so mail delivery, SSO and SAML metadata URLs, payment provider callbacks, monitoring checks, and, with grim regularity, the status page you were planning to update.

Why some users stay online and others see nothing

Resolver behaviour, not attack size, decides what your outage looks like from the outside. That is why the first twenty minutes of the incident call is usually spent arguing about whether there is an incident.

Anyone whose resolver fetched your A record ten minutes ago is still browsing happily until the time to live (TTL) expires. Anyone whose resolver missed is getting SERVFAIL. Negative caching under RFC 2308 keeps NXDOMAIN answers around for the lesser of the SOA minimum and the SOA record’s TTL, with the RFC recommending a cap of three hours, so a bad answer cached early can outlive the attack itself.

Stale-serving adds another layer of inconsistency. RFC 8767 (March 2020) permits resolvers to hand out expired records when the authoritative servers are unreachable, and software such as BIND and Unbound supports it as a configuration option. Some ISPs enable it. Plenty do not. Your customers in one region stay online for hours; your customers on a different network see nothing at all; your own office, whose resolver happens to have warm entries, insists everything is fine.

Third-party names drag you down too. If your checkout loads a payment script from a SaaS hostname whose provider is the one under attack, your zone resolves perfectly and your checkout still fails. Map those dependencies before you need them.

Ten minutes to a working diagnosis: provider, zone or you

Before anyone opens a ticket, establish which of three things is true: the provider’s network is failing, your zone data is wrong, or your delegation has gone.

  1. Query every authoritative nameserver directly. dig @ns1.provider.net yourdomain.co.uk SOA +norecurse, then repeat for each NS in the set. All timing out points at the provider. Some answering and some not points at partial anycast failure, which is still the provider, but changes what you ask them.
  2. Trace from the root. dig +trace yourdomain.co.uk shows whether the parent zone still delegates to the nameservers you expect. A delegation that has vanished is a registrar problem, not an attack.
  3. Check the registration itself. Expiry date, registrar lock status, and whether a well-meant automated renewal failed. Domains still lapse. It looks exactly like an outage.
  4. Validate DNSSEC. dig +dnssec against the provider, and compare the DS record at the parent with the current DNSKEY. An expired RRSIG or a rolled key with a stale DS produces SERVFAIL from every validating resolver while the provider’s network is entirely healthy.
  5. Sample several vantage points. Public resolvers at 8.8.8.8, 1.1.1.1 and 9.9.9.9, plus a phone off Wi-Fi. Anycast means one vantage point tells you almost nothing about global reachability.

Rule out the self-inflicted causes first, because they mimic an attack and they are more common: a zone push that partially applied, a CI pipeline that overwrote the NS set, a CNAME added at the apex, a new provider onboarded but never added to the delegation.

When you escalate, ask closed questions. Which points of presence are affected, and are UK queries served from them? Is the target our zone or a co-tenant? What is the current query success rate by region? Should we prepare a delegation change, and will you support the transfer if we do? An honest provider will confirm within the first hour that they are under attack, even without a vector breakdown. Vague reassurance at hour two is itself information. Worth remembering: providers protect the estate, not the individual tenant. When EveryDNS suffered a sustained DDoS attack in 2010, the resolution was to stop serving the customer whose zone was attracting the traffic.

Dual-provider authoritative DNS: how it works and where it breaks

Two independent operators on one delegation is the only real answer to a provider-level flood. RFC 2182 has recommended diverse secondary nameservers, in different networks and topologies, since 1997. Resolvers already try every nameserver in the NS set, so if four of eight are unreachable, queries still get answered, more slowly.

Synchronisation comes down to two choices. A hidden primary distributing to both providers by AXFR and IXFR zone transfers keeps one source of truth and one place to make changes. API-driven pushes to both providers are easier to start with and drift silently the first time a deploy script only updates one side. Whichever you pick, monitor the SOA serial on both providers and alert when they diverge. That single check catches most of what goes wrong.

Two blockers get skipped in sales calls and discovered mid-incident.

DNSSEC is the first. Two providers each signing your zone with their own keys will produce answers the other cannot vouch for, and validating resolvers will return SERVFAIL. The fix is a multi-signer arrangement under RFC 8901, published in September 2020, where each provider serves the other’s zone-signing keys in the DNSKEY set. Both providers must support it. Many do not, or support it only on their higher tiers.

Vendor-specific record types are the second. ALIAS records, CNAME flattening at the apex, GeoDNS steering, weighted pools and health-check failover are proprietary. They do not replicate through a zone transfer, because they are not records in the standards sense. Failing over to a secondary that has no equivalent gives you a zone that resolves and an application that does not route.

Partial resilience still beats none. A warm secondary, loaded with a current zone and not in the live NS set, plus a written promotion runbook and registrar credentials that work, is a defensible position for a smaller estate. Test it quarterly. Discovering the registrar requires a fax during an outage is a genuine outcome.

Decisions to make before the outage, not during it

TTLs are the clearest example of a choice you cannot revisit once queries are failing. A 300-second TTL means most resolvers return within five minutes of mitigation and gives you almost no ride-through while the attack runs. An 86,400-second TTL keeps existing visitors resolving for a day, then delays any emergency nameserver change by the same day. Set them per record type: longer on NS and MX, moderate on stable A and AAAA, short only where you genuinely need agility, and document a lowering window (drop to 300 seconds at least a full old-TTL ahead) before any planned migration.

Three more decisions belong on the same page:

  • Registrar account access, the NS change procedure and DNSSEC key material held by more than one named person, with multi-factor authentication that does not depend on an email address at the domain you may be about to lose.
  • Status communications on a separate domain, at a separate DNS provider. Your status page on status.yourdomain.co.uk is decoration during this particular failure.
  • A tested breakage order, ranked by revenue: checkout, API clients, mail, SSO. The incident call should prioritise by booked value per hour, not by whoever is loudest on the bridge.

While you have the zone open, audit it for origin exposure. Legacy A records, mail., vpn., dev. and old staging hostnames publish the addresses sitting behind your proxy. An attacker who cannot beat the reverse proxy will happily take the IP your own DNS handed them. It is the same blind spot as origin leakage, dressed differently.

Buying DNS resilience: the evidence to demand and the business case

Authoritative DNS is the surface most often left unmanaged. Teams buy a reverse proxy for the website, arrange BGP scrubbing for their prefixes, sign a sensible contract, and leave the zone on a bundled hosting or registrar nameserver with no capacity claim, no attack reporting and no escalation path. Read your contract and find the words. If the scope stops at your prefixes and HTTP, authoritative DNS is not covered, whatever the cover sheet says. Our guidance on the gaps that leave DNS exposed goes into the specific clauses worth checking.

Trace the resale chain. Several DNS and protection offerings are white-labelled or resold, so two providers can look independent on paper while sharing upstream transit, anycast footprint or scrubbing centres. Ask who owns the network, how many points of presence answer UK queries, and who is actually on the escalation bridge at 03:00. Those questions apply just as much when you are choosing a managed DDoS protection provider in the UK as when you are buying DNS.

Judge capacity claims on evidence, not adjectives: query-per-second headroom per point of presence, redacted post-incident reports from real events, and SLA credit mechanics set against your genuine cost per hour of downtime. Credits are usually a fraction of one month’s fee. Multiply your hourly revenue or booked value by a realistic four-hour DNS incident, compare that to the annual cost of a second authoritative provider, and the procurement case tends to write itself. Judging vendor claims properly is a discipline of its own, covered in our breakdown of how to test six common vendor claims.

One framing does not transfer. The always-on versus on-demand debate is about when you divert traffic into mitigation and how fast. Against a provider-level DNS flood there is nothing to divert and no time-to-mitigate you own. The capacity under pressure belongs to someone else. Delegation diversity across two genuinely independent operators is the equivalent control, and it is bought in advance or not at all.

Dyn remains the reference case for concentration risk. Three waves of attack traffic in October 2016, driven by Mirai-infected consumer devices, took well-engineered sites offline while their own infrastructure ran normally. Dyn’s own post-incident analysis, published within days, described tens of millions of discrete IP addresses associated with the botnet and pointedly declined to confirm the 1.2Tbps figure circulating in the press. The scale was never the lesson. The dependency was.

Frequently Asked Questions

Does a second DNS provider really help when one is under attack?

Yes, provided the two are genuinely independent and both are live in the NS set. Resolvers try each nameserver listed, so queries still resolve when half the set is unreachable, though users may see added latency. The benefit disappears if both providers share upstream transit, anycast infrastructure or the same parent company.

Will lowering TTLs protect my site during a DNS provider outage attack?

No, and it can make things worse. Short TTLs mean caches empty faster, so more users hit the failing authoritative servers sooner. Low TTLs help you move quickly during a planned migration or a nameserver change; longer TTLs give more ride-through during an outage. Either way, changing them after queries start failing has no effect, because the change has to propagate through the servers that are already unreachable.

Is authoritative DNS covered by a standard managed DDoS protection contract?

Often not. Many contracts scope cover to your IP prefixes and to HTTP traffic through a reverse proxy, which excludes a zone hosted on a third party’s nameservers entirely. Check the service description for an explicit reference to authoritative DNS, query-per-second protection and DNS-specific attack reporting, and treat silence as exclusion.

How long does recovery take once the DNS provider mitigates the attack?

Longer than the mitigation itself. Cached negative answers persist for their full negative TTL (capped under RFC 2308 guidance at three hours), resolution failures may be cached for up to five minutes under RFC 9520 guidance, and resolver retry backlogs take time to clear. Expect a tail of inconsistent user reports for a period roughly matching your negative caching settings, not a clean switch back on.

Can I run DNSSEC across two DNS providers?

Only with a multi-signer setup as described in RFC 8901, where each provider publishes the other’s zone-signing keys so answers from either validate correctly. Both providers must support the model and be willing to operate it together. Without it, one provider’s signed answers will be rejected as bogus by validating resolvers, which turns your resilience investment into an outage of its own.

DNS DDOS Attack Protection: 6 Gaps to Close

A retailer can spend two years and a five-figure annual contract getting its web estate right, with every hostname behind a reverse proxy, always-on scrubbing across the origin range and an escalation matrix taped to the wall of the NOC, and still go completely dark because the nameservers answering for its domain arrived free with the domain registration. DNS DDoS attack protection is the surface almost nobody in the org chart actually owns. The proxy contract has a signatory. The scrubbing centre has an account manager. Authoritative DNS usually has a login someone set up in 2019.

What follows is a review of your own setup rather than a product explainer: six specific gaps, the deployment models that close them, the contract questions that separate real capacity from resale, and an audit you can run this week.

Why authoritative DNS is the protection surface most contracts quietly skip

Distributed denial-of-service (DDoS) protection is worth assessing across three distinct surfaces, not one. Network and transport (volumetric floods, SYN floods, UDP reflection aimed at your IP ranges). Application layer (HTTP floods, expensive search queries, credential-stuffing traffic that looks like customers). And authoritative DNS, which is the system that answers the question “where is www.yourdomain.co.uk?”for every recursive resolver on the internet.

Buyers routinely assume the third is covered by the first two. It rarely is, for a mundane reason: DNS sits with the registrar or the hosting control panel, not the security vendor. A content delivery network protects the traffic that reaches its edge. If your zone is served by four nameservers on a shared platform bundled with a reseller hosting account, nothing in your scrubbing contract touches them.

How the zone ends up unmanaged

Three patterns account for most of it. The domain was registered years ago and the registrar’s free nameservers were never changed, because they worked. DNS was set up in a cPanel or Plesk instance and inherited by whoever bought the hosting company. Or there is a legacy secondary nameserver from a previous supplier still listed in the delegation, on a box nobody has logged into since the last account manager left.

None of these are negligence exactly. DNS is the one part of the stack that keeps working with no attention at all, right up until it does not.

The reachability chain, and where a single weak link ends it

User to resolver, resolver to your authoritative nameserver, client to the proxy edge, edge to origin. Every one of those hops has to work. Break the second and the other three are irrelevant, which is what makes the failure so slow to diagnose: origin monitoring stays green, the scrubbing dashboard shows a healthy baseline, the edge reports no anomaly, and the phones start ringing anyway. Synthetic checks that resolve once and cache the answer will lie to you for the length of their TTL. If you want the mechanics of that failure at length, our piece on DDoS on DNS and how blackouts happen covers it.

The attack types your DNS protection has to survive

Direct query floods. Straightforward volumetric pressure against the IPs of your authoritative servers, in packets per second rather than gigabits. DNS servers fall over on query rate long before the pipe fills.

Pseudo-random subdomain attacks, often called water torture. The attacker generates names like a8f3k2.yourdomain.co.uk, which have never existed and cannot be in any cache. Recursive resolvers worldwide have no choice but to forward every one of them to you. The global resolver population becomes an unwitting amplifier, and the traffic arrives from thousands of legitimate resolver IPs you cannot simply block without blocking real customers behind them.

Reflection and amplification against your zone. Spoofed queries with your victim’s address as the source, aimed at open resolvers, requesting responses far larger than the query. ANY queries were the classic vehicle, which is why the IETF published RFC 8482 in January 2019, standardising minimal-sized responses to QTYPE=ANY. Note the direction of the harm here: your zone is the ammunition, not the target. DNSSEC-signed zones return larger answers by design, which raises the amplification factor available to whoever points a reflector at them. Preventing your own resolvers from being used this way is the subject of BCP 140.

TCP-based DNS floods. Cheaper to defend at the network layer, but they exhaust connection state on nameservers tuned for UDP, and truncation behaviour means legitimate large responses need TCP to work.

Collateral damage. On a shared DNS platform, an attack aimed at another customer’s zone lands on the same anycast prefixes serving yours. In December 2010, EveryDNS terminated DNS service for wikileaks.org, stating that the attacks directed at that domain threatened the stability of the infrastructure serving its other customers. The 21 October 2016 attack on Dyn, which Dyn confirmed involved Mirai botnet traffic, took down name resolution for a long list of well-known services that had done nothing to attract attention. Your risk on a shared platform is not only your own threat profile. It is everyone else’s.

Six gaps to close in your DNS DDoS attack protection

Gap 1: one authoritative provider, no genuinely independent secondary

Two nameservers on the same anycast network, in the same account, sharing the same control plane, are one nameserver with a redundant hostname. Provider-wide events are the ones that hurt, and they are not rare. A real secondary means a different company, a different anycast footprint, separate BGP announcements and separate billing. The test question: if this provider’s entire platform disappeared for 90 minutes, would your domain still resolve? If the answer needs a caveat, you have one provider.

Gap 2: marketing uptime figures instead of capacity evidence

“100% DNS uptime SLA”is a credit policy, not a capacity statement. Ask for queries per second the platform sustains, how many anycast points of presence serve UK and Western European traffic, whether those PoPs are owned or leased, and what happens to a single PoP under load (does it withdraw the route and shift traffic to the next site, and what is the resulting latency for a resolver in London?). Rate limiting matters too: response rate limiting, implemented in BIND since 9.9.4, will drop repeated identical responses, but it does nothing against randomised subdomains because no two responses are identical. Ask specifically what they do about that vector.

Gap 3: TTLs set for convenience rather than failover

A record at 86400 seconds means a failover decision takes up to a day to reach most of the resolver population. At 300 seconds, emergency re-pointing is realistic within minutes, at the cost of higher query volume against your nameservers and a marginally higher bill on per-query pricing. Most revenue-carrying records belong at 300 to 900 seconds as a standing posture, not lowered in a panic during the incident (by then it is too late, the old value is already cached).

The subtler trap is negative caching. Under RFC 2308, how long a resolver remembers an NXDOMAIN is governed by the SOA MINIMUM field and the SOA record’s own TTL, so an SOA minimum of 86400 can leave “this name does not exist”cached long after your zone is healthy again. And TLD delegation NS records commonly carry TTLs measured in days, which means switching providers at the registrar is a planned migration, never an emergency lever.

Gap 4: registrar and zone accounts with no MFA and stale delegation hygiene

A hijacked registrar account achieves everything a DDoS does, faster and quieter. Multi-factor authentication on the registrar, registry lock where the registry supports it, and a role account rather than a departed engineer’s personal email. Then check the delegation itself: do the NS records in your zone match the NS set at the parent, are the glue records pointing at IPs you still control, and is there a fifth nameserver in the list that has not answered since 2021? Stale glue is not a theoretical problem. It is an IP someone else can eventually be allocated.

Gap 5: nobody named as the person who acts during a DNS incident

The dividing line is simple. If no human on the provider’s side is empowered to change a rate limit or push a response policy at 03:00 without a change board, you have bought monitoring, not management. Same test for your side: name the person who can lower a TTL, add a secondary, or re-point a record out of hours, and confirm they hold the credentials to do it. Our first-60-minutes playbook for ecommerce incidents assumes that person exists. Frequently they do not.

Gap 6: the origin IP is still published somewhere in your DNS

This is the one that undoes the entire proxy investment. The leak almost always lives in DNS: an mail.yourdomain A record pointing straight at the same server the website runs on, an ftp or webmail host auto-created by a control panel, an SPF record with the origin’s IP in a mechanism, a dev or staging name that was never proxied. Historic passive DNS and certificate transparency logs preserve the rest. Attackers do not need to beat your edge if a five-year-old record hands them the address behind it. Enumerate every record in every zone, and treat any that resolves to origin as an open door.

Deployment models, and where each one fits

Managed anycast DNS from a specialist. Absorption capacity you cannot replicate in-house, plus operational responsibility for the query layer. You give up control of tuning and you inherit the provider’s blast radius when someone else’s zone attracts a botnet.

Self-hosted nameservers behind BGP scrubbing. Viable if you run your own AS and have the on-call depth to support it. The diversion trade-offs are the same as for web traffic, and worth reading against the always-on versus on-demand comparison. The consequence differs, though: a few minutes of unanswered queries outlast the attack itself, because resolvers cache the failure and clients retry unevenly. Web traffic recovers when the flood stops. DNS recovers when the caches do.

Registrar or hosting-panel DNS. Fine for a parked brand-defence domain or an internal tool. Rarely defensible for a domain that carries revenue, because you have no capacity evidence, no incident contact and no ability to influence mitigation.

Multi-provider DNS with AXFR/IXFR or API-based sync. The pattern that actually survives a provider-wide event: two independent platforms, both authoritative, both listed in the delegation. The operational cost is real. Zone drift between platforms, DNSSEC key management across two vendors (usually solved by keeping signing with one and transferring signed data, or running multi-signer per RFC 8901), and a change process that pushes to both. Most organisations that skip this cite complexity. Most that adopt it do so the week after an outage.

How to verify a provider’s claims before you sign

A great deal of DDoS and DNS capacity is resold. Ask explicitly whose anycast network it is, whose scrubbing centres handle the traffic, how many points of presence serve UK queries, and whether the party on your contract can change mitigation policy directly or can only raise a ticket with the upstream. The answer to that last question determines your real time to mitigate.

“Managed”is a contract word, not a technical one. For DNS it should name who performs zone onboarding, who tunes rate limits, who is authorised to push a response policy or TTL change mid-incident, and what the out-of-hours escalation path is by name and number. If nobody is named, the answer during an attack is you.

Read the service level agreement for what it measures rather than what it promises. Is authoritative DNS covered by the same time-to-mitigate commitment as web traffic, or is it excluded in a schedule? Is the clock started by your ticket or by their detection? The arithmetic on service credits is usually unflattering once you compare it with an hour of lost trading, and what those SLA seconds actually represent deserves a closer look than the sales deck gives it. Then ask for a redacted post-incident report from a real DNS event: vector breakdown, query rates, time to detection, actions taken. A dashboard screenshot is not evidence.

A DNS resilience review you can run this week

  1. Inventory. Every domain the business owns, including the ones marketing registered for a campaign; who is authoritative for each; who holds the registrar login; and the expiry date. Expiry is not a DDoS problem, but it takes sites down just as effectively.
  2. Delegation and record audit. Compare the parent NS set with the zone’s own; check glue; list every A and AAAA record and flag anything resolving to origin; check MX and SPF for leaked addresses; record current TTLs and the SOA minimum; confirm DNSSEC status and who holds the keys.
  3. Failover rehearsal. Write down what you would change, who has permission, and how long propagation genuinely takes at today’s TTL values. Rehearse it on a low-stakes domain. Most teams discover the credentials problem before they discover the technical one.
  4. Business case. Downtime cost per hour against the annual cost of a secondary DNS provider or a managed contract. UK pricing is generally structured around clean bandwidth or protected IP ranges for network mitigation, with DNS charged per zone and per million queries; our guide to how managed DDoS protection is priced in the UK sets out the variables to compare. Secondary DNS is usually the cheapest resilience you will buy all year.

Sound dns ddos attack protection is less about buying a bigger scrubbing pipe than about removing the assumption that someone else already covers the query layer. Check who answers for your domain, and check tonight rather than during the incident.

Frequently Asked Questions

What is a DNS DDoS attack and how does it differ from a website flood?

A DNS DDoS attack targets the nameservers that translate your domain into an IP address, rather than the web servers that return pages. The difference matters operationally: during a web flood the site is slow or erroring, while during a DNS attack the site is simply not found, and origin monitoring can stay green throughout because the origin is fine. Nobody can reach it.

Does my CDN or scrubbing provider already cover authoritative DNS?

Sometimes, but never assume it. Many CDN plans include managed DNS as an option that must be enabled and delegated to, and plenty of customers proxy their web traffic while leaving the zone with the registrar. Check the NS records for your domain right now; whoever appears there is the party responsible, whatever the security contract says.

Do I need a second DNS provider, or is anycast enough on its own?

Anycast handles volumetric absorption well and is the right baseline. It does not protect you from a provider-wide control plane failure, a routing mistake, a billing suspension or a sustained attack on a neighbouring customer. For revenue-carrying domains, two independent providers is the pattern that survives those events.

Does DNSSEC protect against DDoS attacks?

No. DNSSEC authenticates responses and defends against cache poisoning and spoofing; it has no bearing on whether your nameservers can answer under load. Signed responses are larger than unsigned ones, which can increase the amplification factor available in a reflection attack that uses your zone, so DNSSEC and capacity planning need to be considered together rather than treated as the same control.

What TTL values make DNS failover realistic during an attack?

For records you might need to move in a hurry, 300 to 900 seconds is a workable standing value; at 86400 seconds a failover decision can take most of a day to reach the resolver population. Lower the TTL well before you need it, since the old value is already cached by the time an incident starts, and check the SOA minimum as well, because negative caching under RFC 2308 governs how long an NXDOMAIN sticks around.