DDOS Protected Hosting: What You Actually Get

DDOS Protected Hosting: What You Actually Get

“DDoS protected”sits on a hosting quote the way “secure”sits on everything else: two words doing a great deal of unexamined work. DDoS protected hosting can mean a shared account sitting behind an upstream filter the host neither owns nor operates, or it can mean a dedicated server with a per-IP scrubbing profile, a documented time-to-mitigate and an engineer who answers the phone at three in the morning. Same label. Same price bracket, sometimes. Very different outcomes when the junk traffic arrives.

What follows is a way of decoding the claim rather than a league table of providers. The test applied here is the same one used to assess any protection story: three surfaces, not one. Network and transport (layers 3 and 4), the application layer (layer 7), and authoritative DNS. Hosting plans almost always address the first. Some gesture at the second. Almost none touch the third, which is the surface that makes every other defence irrelevant the moment it fails.

What “DDoS protected hosting”means on a typical invoice

The phrase covers a spectrum that runs from near-nothing to genuinely useful:

  • Shared hosting on a protected range: an upstream filter at the data centre edge, shared by every tenant, with no per-site configuration and usually no reporting;
  • A VPS inside a “protected”IP block: the same perimeter, plus whatever the hypervisor host chooses to enforce on noisy neighbours;
  • A dedicated server with a per-IP scrubbing profile: thresholds set for your address, sometimes tunable, sometimes not;
  • Colocation with a BGP diversion option: your prefixes, your announcement, and a diversion path to a scrubbing centre that you or the provider trigger.

In the first two cases, the label describes the data centre’s perimeter, not a service bought on your behalf. The host protects its own network because an attack on one customer degrades everybody. You benefit as a side effect. That distinction matters because the quiet default in many plans is that protection stops at exactly the point your traffic starts threatening other tenants, and at that point the cheapest remedy available to the host is to remove you from the internet.

Null-routing: the clause that decides whether your site stays up

Blackholing, or remote triggered black hole filtering, is the oldest and bluntest tool in the box. The technique is documented in IETF standards going back to RFC 5635 and its predecessor RFC 3882, and the mechanics are simple: the host tags your destination address with a community that tells its upstream providers to discard every packet addressed to it. Attack traffic stops. So does your traffic. The network survives, the customer does not.

It is cheap for the host, instant, and total for you. It is also, in most acceptable use policies, entirely permitted. A provider that blackholes your IP at 1Gbps of inbound attack traffic and leaves it null-routed for two hours has technically honoured a “DDoS protected”claim while your site has been unreachable for the whole incident.

Read the acceptable use policy before the sales page. Four things decide how much the protection is worth:

  • The threshold at which null-routing is triggered, expressed in bandwidth or packets per second;
  • The minimum blackhole duration, which is frequently fixed at one, two or twenty-four hours regardless of when the attack stops;
  • The escalation path for repeat attacks, including clauses that allow suspension or termination after a set number of incidents in a billing period;
  • Who can lift the null route, and whether that requires a ticket, a phone call or an engineer on a shift you are not paying for.

Shared addressing makes this worse. On a shared IP, a neighbour’s attack is your outage: the upstream drops traffic to the address, and every site on it goes dark together, with no fault on your side and no visibility into what happened. A dedicated IP changes your exposure before a single filtering rule is applied. That alone is often the most valuable line item on a protected hosting invoice.

Reading the capacity and mitigation claims on a hosting page

Headline capacity figures are the most misread numbers in the industry. A host advertising several Tbps of protection is describing the aggregate network capacity of its data centres, or the total capacity of the upstream mitigation partner it buys from, across every customer on the platform. It is not an allocation reserved for your plan, and nothing in the contract usually says it is. Cloudflare reported in June 2025 that it had mitigated an attack peaking at 7.3 Tbps, which gives some sense of how quickly aggregate headroom becomes a shared resource under pressure.

Two questions get past the marketing. What is the per-IP or per-port filtering cap, which on many platforms sits far below the headline figure? And what happens when an attack exceeds it: does traffic get rate-limited, does it get diverted upstream, or does the address get null-routed? The answer to the second question is the one that matters, and it is rarely on the sales page. There is a fuller treatment of how those numbers are constructed in this breakdown of how much scrubbing capacity you actually need.

Always-on filtering versus on-demand diversion

These are not interchangeable, and for hosting customers the difference is measured in minutes of downtime. On-demand means detection, then a routing change, then mitigation. Each step takes time, and the gap between them is the window in which your site is down. Always-on removes the gap because traffic is inspected continuously, at the cost of permanent inspection and, on some stacks, a small latency penalty that matters to latency-sensitive applications and almost nobody else.

Ask where scrubbing physically happens, who operates it, and whether the kit belongs to the host or to a third party it resells. A provider that answers “our firewall drops SYN floods”is describing a firewall rule, not a mitigation stack. The same line of questioning applies to upstream carriers, and the six questions worth putting to an ISP about its DDoS protection transfer almost unchanged to a hosting provider.

The application layer and DNS gaps most hosting plans leave open

A volumetric filter is tuned to recognise floods: spoofed UDP, reflected amplification, SYN storms, traffic that looks nothing like a browser. A slow HTTP flood against a login form, a search endpoint or a basket page looks exactly like a busy afternoon.

The scenario repeats constantly. An ecommerce site on a protected VPS absorbs a layer 3 flood without the operator noticing, which is exactly what was paid for. A fortnight later, a few thousand requests per second against its search and basket endpoints put the database on the floor, because every one of those requests is well-formed HTTP from a residential IP range and the volumetric filter reads it as legitimate. No alert fires. The host’s dashboard shows no attack.

A hosting control panel can usually do a few useful things here: edge caching for anonymous traffic, basic rate limiting per IP, a web application firewall with managed rulesets if the plan includes one, and a challenge page on request. What it generally cannot do is distinguish a distributed low-rate flood from a flash crowd, apply behavioural scoring, or protect the uncacheable endpoints where the damage happens. That needs a separate layer, and it needs to be configured before the incident, not during it.

Authoritative DNS, the surface nobody manages

Here is the gap that embarrasses otherwise careful setups. A domain’s nameservers stay on the registrar’s free bundled DNS because that is where they were pointed on day one, while the web tier sits behind a well-mitigated host. The web tier is defended. The domain still stops resolving when the nameservers are hit, and a site nobody can resolve is a site nobody can reach. Every other control in the stack is downstream of that lookup.

Check who runs your authoritative DNS today, not who you think runs it. Run a lookup for the NS records of your own domain, count how many distinct providers appear, and ask whether that provider publishes anything at all about anycast coverage, query capacity or response rate limiting. Registrar-bundled DNS is usually fine until it is targeted. DNS infrastructure is also attractive as an attack tool rather than a target, which is a separate problem covered in this guide to what actually works against DNS amplification attacks.

The real cost, including what lands on the invoice afterwards

UK managed DDoS pricing is structured in layers, which is why the monthly headline so rarely matches the invoice. The usual components are a protected IP fee, clean-bandwidth or overage billing on the traffic that survives filtering, a tiered uplift on dedicated servers, and a managed add-on if you want somebody else to tune the profile.

Then there are the charges that only appear after an incident: burst traffic billed at the 95th percentile, emergency onboarding if you arrive mid-attack, out-of-hours engineering time, and contract terms triggered by repeat attacks. Ask for a worked example of a billed incident before signing. A provider that cannot produce one has not thought about it either.

Build the comparison from downtime, not from the monthly line item. An operator who can state a per-hour revenue figure for an outage (orders per hour times average order value, plus support cost, plus any SLA credits owed to their own customers) has a number to judge every quote against. Without it, you are comparing £ against £ with no sense of what either buys. The same arithmetic decides when an uplift is money well spent and when the budget belongs in a separate mitigation service instead. Related reading on where ISP-level protection runs out and when to top it up covers that threshold in more detail.

Protected hosting, a scrubbing service, or both

Risk profile decides this, not budget. Public-sector adjacency, hacktivist attention, gambling and crypto exposure, retail peak trading and any history of extortion demands all move a business out of the category that hosting-level protection serves well. The UK’s forthcoming Cyber Security and Resilience Bill adds another consideration for suppliers to regulated sectors, where availability expectations are becoming contractual rather than aspirational.

For brochure sites, internal tools and low-value targets, hosting-level protection is a reasonable floor and the money is better spent elsewhere. Where downtime is billed in lost orders, dedicated mitigation with a defined time-to-mitigate earns its keep. If you buy both, keep the paths independent: hosting protection, authoritative DNS and the mitigation provider should not collapse into a single vendor or a single point of failure, because the attack that takes out one will be the attack that matters.

Before signing, ask for a sample mitigation report from a real incident with the customer details removed, the contact route to use during an attack (a named number, not a web form), the documented escalation times, and written confirmation of the null-route threshold and duration. The NCSC’s denial of service guidance collection is a reasonable neutral reference for what a provider should be able to answer. If the sales team cannot supply those four things in writing, the protection you are buying is a label.

Frequently Asked Questions

Is DDoS protected hosting enough on its own?

For low-value targets, often yes. For anything where an hour of downtime has a revenue figure attached, rarely. Hosting protection typically covers network and transport floods at the data centre perimeter, leaves application-layer attacks largely untouched and ignores authoritative DNS completely, so the honest answer depends on which of those three surfaces your attacker chooses.

What does null-routing mean and will my host do it to me?

Null-routing, or blackholing, tells upstream routers to discard all traffic destined for your IP address. Attack traffic stops and so does every legitimate visitor. Most hosting acceptable use policies permit it above a stated threshold, so assume your host will use it unless the contract says otherwise, and get the threshold, the minimum duration and the process for lifting it confirmed in writing.

Does DDoS protected hosting cover application layer attacks?

Usually not in any meaningful way. Volumetric filtering is tuned to recognise floods, and a few thousand well-formed requests per second against a login, search or basket endpoint reads as ordinary HTTP traffic. Caching, rate limiting and a web application firewall help, but distinguishing a distributed low-rate flood from genuine demand generally requires a layer the hosting plan does not include.

How much more does hosting with DDoS protection cost in the UK?

Pricing is layered rather than flat: a protected IP fee, clean-bandwidth or overage billing, a tier uplift on dedicated servers and optional managed configuration, plus post-incident charges such as burst traffic and out-of-hours engineering. Figures move constantly, so ask each provider for a worked example of a billed attack and compare the total against your own downtime cost per hour.

Does my hosting plan protect my authoritative DNS?

Almost certainly not. Authoritative DNS usually sits with the registrar or a separate provider and is outside the hosting contract entirely, which is why it is the surface most often left unmanaged. Check your domain’s current NS records, see how many distinct providers appear, and ask that provider directly what its anycast footprint and query capacity look like.

The decoding framework above is the useful part of any DDoS protected hosting conversation: three surfaces, the null-route clause, the per-IP cap behind the headline number, and the invoice that arrives after the attack. If you are working through a quote now, the companion piece on the limits of ISP-level protection is the next thing worth reading.