Announcing a capacity proves nothing. Here is what our filtering servers actually handled, measured by their own telemetry, with the counting method and the limits of the exercise.
Measurement period : → (95 days)
Hourly distribution of detections over the period. Not one of the twenty-four hours is empty — that is what makes continuous detection credible, rather than a claim.
Identified vectors, as a share of qualified attacks. Attacks whose vector could not be qualified are excluded from this calculation, and their number is stated.
5 267 attacks whose vector was not qualified.
The largest volumetrics of the period. No prefix, no address, no customer name: only the size of the targeted block is shown, because it carries technical meaning.
| Detected | Packet peak | Bandwidth peak | Vector | Duration | Target block | State |
|---|---|---|---|---|---|---|
| 2026-08-20 15:04 | 934 Mpps | 598 Gbps | unknown | 485 min 52 s | — | Resolved |
| 2026-08-23 10:00 | 905 Mpps | 2.81 Tbps | unknown | 18 min 45 s | — | Resolved |
| 2026-08-24 21:59 | 848 Mpps | 542 Gbps | unknown | 695 min 39 s | — | Resolved |
| 2026-08-24 21:59 | 816 Mpps | 522 Gbps | unknown | 695 min 39 s | — | Resolved |
| 2026-08-25 11:31 | 808 Mpps | 517 Gbps | unknown | 545 min 38 s | — | Resolved |
| 2026-08-25 11:31 | 776 Mpps | 496 Gbps | unknown | 545 min 38 s | — | Resolved |
| 2026-08-20 15:04 | 772 Mpps | 494 Gbps | unknown | 485 min 52 s | — | Resolved |
| 2026-08-20 12:00 | 678 Mpps | 434 Gbps | unknown | 172 min 57 s | — | Resolved |
| 2026-08-24 21:59 | 670 Mpps | 429 Gbps | unknown | 695 min 39 s | — | Resolved |
| 2026-08-24 06:15 | 555 Mpps | 14 Mbps | unknown | 31 min 37 s | — | Resolved |
An attack is an episode recorded by our filtering servers’ telemetry: detection timestamp, packets-per-second peak, bandwidth peak, vector where it can be qualified, and resolution timestamp. The peak reported is the maximum observed over the episode, not an average.
Isolated outliers. A single reading more than three times above the 99.9th percentile is not treated as evidence until confirmed: it is more likely a measurement anomaly. One such reading exists in this period and does not appear in these figures.
No prefix, no IP address, no customer or service identifier. The size of the targeted block is kept because it tells you about the nature of the attack without identifying anyone. No customer name will appear here without written permission.
These figures describe what our servers saw and handled. They do not measure a false-positive rate, nor end-user perceived quality, nor a time to mitigate: the mitigation start time is only timestamped on a small share of episodes, so we do not publish it. We would rather say so than average a non-representative sample.
