CorePilot DDoS · provided by Direct Communications

DDoS Protection — Terms of Service

What the service does, what it aims for, and — stated plainly and up front — what it does not and cannot promise.

← Back to sign in
01

The service

Traffic crossing the transit edge toward your address space is monitored for volumetric denial-of-service attacks. When one is identified, mitigation is applied at the network edge, you are notified, and the current state and history of your circuits are visible in the customer portal.

Protection is provided per circuit — a location, its registered address ranges, and the contracted rate agreed for it — for a stated term.

Where a plan is described as covering unlimited ports, addresses or ranges, that refers to how much you may register, not to how much attack traffic can be absorbed. Every circuit is judged against its own contracted rate, and the capacity of the edge and of the transit circuits behind it is finite and shared with every other customer. No plan, however described, can absorb an attack larger than the circuit carrying it — see section 02.

A circuit with no term recorded is watched, not defended. Attacks on it are still detected, recorded, notified and shown to you in the portal — what is withheld is the automatic action. The portal shows such a circuit as monitored rather than protected, and the two words mean exactly that difference. An operator can still act by hand at any time.

Where the address space belongs to a network that buys transit from us rather than to one of our own subscribers, automatic action additionally requires the authorisation of the network that holds the relationship with it. Without that authorisation we will not act on that space, and with it an operator still approves each action by hand before it is applied, because dropping another network's customer traffic is not a decision to automate.

The shape of what is offered

This is a best-effort mitigation service, not an availability guarantee. We commit to watching, to acting, and to telling you what we saw. We do not commit to any particular outcome of an attack, because the outcome depends on the size and shape of the attack and on the capacity of networks upstream of ours.

02

Why this is best effort

These reasons are specific and measured. They are stated here rather than buried in a liability clause, because knowing them lets you make better decisions about your own resilience.

An attack bigger than the pipe cannot be fixed at the far end of it

Mitigation is applied at our edge routers. An attack large enough to saturate a transit circuit has already congested that circuit before the packets reach a router we can program. Dropping them at our edge does not un-congest the link they arrived on. This is the most important limit, and it is a property of physics and topology rather than of any product.

Detection takes minutes, and some attacks are shorter than that

Flow is exported by the routers, written into 60-second capture files, and evaluated once a rotation. A finding must persist across two consecutive captures before mitigation may act — a single measurement artifact must never disconnect a customer. That is roughly two to four minutes for a sustained attack. An attack shorter than that is recorded and reported, and is over before mitigation could act.

Measurement is sampled

Flow is sampled rather than counted packet by packet — commonly somewhere between 1:200 and 1:1000, set per network. Small or low-rate attacks may not be visible, and every bandwidth figure shown is a statistical estimate rather than an exact count.

Automatic action covers specific attack shapes

Automatic mitigation is limited to signatures that can be identified with confidence. Other classes — including application-layer attacks, low-and-slow attacks, and attacks that resemble legitimate traffic — are reported but not automatically mitigated. The restraint is deliberate: acting on a weaker signal disconnects customers who are not under attack.

Mitigation is a blunt rule, not a scrubber

Mitigation matches a class of traffic at the edge and either rate-limits it or discards it, depending on how the network is configured. It does not distinguish an attacker's packet from a legitimate packet of the same shape, so some legitimate traffic in that class is affected while a rule is in force. This service does not include traffic scrubbing or packet-level cleaning.

If an attack keeps changing shape, we stop delivering the address

Mitigation escalates. A single steady attack is answered with a narrow rule covering one port. If the attack moves to another port, the rule widens to the whole protocol. If it keeps moving — several distinct vectors, a change of protocol, or roughly twenty minutes sustained — the final step is to stop delivering the attacked address at our edge altogether.

This is the strongest action available to us, and it is worth being plain about what it means for you: while it is in force that address is unreachable for everyone, not only for the attacker. From your side the result is hard to tell apart from the attack having succeeded. We choose it because the alternative is a flood that keeps arriving and degrades the shared infrastructure every other customer depends on, and because a narrower rule has already been tried and outrun.

It is withheld in one case: where the evidence for the attack came only from your own circuit, discarding the address would complete the attack rather than stop it, so the narrower rule is kept instead.

It depends on systems no single party controls

Detection depends on the routers exporting flow and on the collectors receiving it. Mitigation depends on the routers installing the rule. Upstream transit providers may fail, reroute or become congested independently of us.

What follows from all of this

These limits are stated so you can plan around them rather than discover them during an attack. Treat this service as one layer and not as the whole answer. If your business cannot tolerate being unreachable during a large attack, the measures that help are ones we cannot provide from our edge: headroom on your own circuit, addresses that can be moved or re-announced, and — for the largest attacks — a scrubbing arrangement upstream of us. We would rather tell you that now than after the event.

03

What is committed, and what is a target

The clearest thing we can do is separate what we commit to from what we aim for and what we will never promise.

TierCovers
CommittedMonitoring is running. You are notified. Records are kept and visible. Mitigation is withdrawn when the attack stops. A person responds — this one is a commitment of the operator running this service, not a property of the platform.
TargetedTime to detect, time to mitigate, and which attack classes are handled automatically.
Never promisedThat your service stays up during an attack. That a given volume is absorbed. That every attack is detected. That no legitimate traffic is affected.
Before the numbers below

The table that follows is a list of targets, and a reader who stops at it would be reading numbers that look like commitments. They are not. We do not promise that your service stays up during an attack, that any given volume is absorbed, that every attack is detected, or that no legitimate traffic is affected. Nothing in the table changes that, and a target that is missed is not a breach of this agreement.

The targets below describe how the platform is configured and what it has been measured to do. They are targets, not warranties.

Targets, not warranties — a missed target is not a breach.
MeasureTarget
Detect a sustained qualifying attack< 5 min
Apply automatic mitigation< 5 min
Notify you< 15 min
Withdraw mitigation after traffic clears< 10 min
Maximum unattended mitigation hold60 min
Attack history in the portal12 months
04

What is not included

05

Your obligations

06

Acceptable use

07

What we may do, and when

So that there is no doubt about the authority being exercised:

08

Information we hold

Detecting attacks means recording traffic, and traffic records are addresses. What is kept is listed here rather than described in general terms, because a general description is not something you can check.

Where it stays

All of it is held on the infrastructure of the operator providing this service. None of it is sent to the software vendor — the platform verifies its own licence locally and has no channel back. In the portal you see your own circuits and nothing about any other customer; what is withheld from you is other subscribers' traffic and the internal detail of the network carrying it.

These records exist to detect attacks, to act on them, and to show you afterwards what happened. Ask and we will tell you what is held for your ranges.

09

Events outside our control

Section 02 explains these limits as a matter of physics and topology. This is the same list stated as obligation: the targets in section 03 do not apply, and no failure of this agreement arises, where the cause is

10

Term, renewal and cancellation

11

Limitation

Reasonable endeavours

The service is provided on a reasonable-endeavours basis. We do not warrant that any attack will be detected, prevented or successfully mitigated, nor that your service will remain available during an attack.

We are not liable for loss arising from an attack, from the effects of mitigation applied under these terms, or from the failure of any upstream network.

Where liability cannot be excluded, our total liability for all claims relating to a circuit is limited to the charges paid for that circuit in the twelve months before the claim, and no service credits are payable unless the commercial agreement expressly states them. We are not liable in any case for loss of profit, revenue, data or goodwill, or for indirect or consequential loss.

Nothing here limits liability for death or personal injury caused by negligence, for fraud, or for anything else that cannot lawfully be limited.

12

Definitions

Circuit
A location, its registered address ranges, and its contracted rate. The unit protection is provided against.
Contracted rate
The bandwidth agreed for a circuit. Attacks are judged against this rather than against the size of the whole edge — a 1 Gbps customer knocked offline is invisible as a share of a 120 Gbps edge and obvious as a share of their own circuit.
Qualifying attack
An attack carrying a signature the platform identifies, persisting across two consecutive one-minute captures, and meeting the rate thresholds in force.
Mitigation
A rule applied at the network edge that rate-limits or discards the matching traffic class toward the affected address, and at its final step stops delivering that address altogether.
Best effort
The monitoring, detection and mitigation described here, applied using reasonable endeavours, without warranting any outcome.