Compliance
Data residency, explained for growing teams
Where your data lives, what backups change, and the questions worth asking.
Security
In the second quarter we mitigated fourteen attacks above 100 Gbps. The largest peaked at 410 Gbps and lasted nine minutes. No customer saw packet loss for longer than the time it took BGP to converge on the scrubbed path. This is a short review of what showed up and how the network behaved.
Scrubbing is always on. When traffic to a protected prefix crosses a threshold or matches an attack signature, we advertise the prefix to the scrubbing cluster via BGP. The cluster filters, rate-limits and returns clean traffic to the origin. The diversion is automatic; nobody needs to call us at 02:00 to ask for protection.
We do not null-route a customer's IP unless they ask us to, or unless the attack threatens the platform itself. Null-routing is a last resort, not a first response: it stops the attack by stopping the service.
The mitigations worked because the capacity was tested in March, the runbook had been updated in February, and the on-call engineer had seen the graphs before. The most valuable work happened when nothing was attacking us.
We publish a monthly summary of attack volume and types to customers. If you would like the Q2 numbers in more detail, ask support. If you would like to argue about scrubbing architecture, ask Priya. She will probably draw a diagram.