The hidden cost of retries during a flood (part 4)
Over the last few months we have spent a lot of time on the hidden cost of retries during a flood. This is what we learned.Latency budgetOur budget for the whole filtering…
Over the last few months we have spent a lot of time on the hidden cost of retries during a flood. This is what we learned.Latency budgetOur budget for the whole filtering…
login endpoint flood against a developer platform platform in Iberia, peaking at 908.2 million requests per second. Mitigated in 0.193 seconds.
cache-busting query flood against a sports betting platform in the Middle East, peaking at 231.2 million requests per second. Mitigated in 0.905 seconds.
carpet-bombing UDP flood against a electronics retail platform in Iberia, peaking at 351.9 Gbps. Mitigated in 0.118 seconds.
This post is about per-tenant rate limits for SaaS platforms. It started, as most of our posts do, with an incident that did not go the way we expected.Lessons for your own…
Over the last few months we have spent a lot of time on hTTP/2 rapid reset one year later. This is what we learned.Testing in production, safelyEvery rule starts in log mode.…
cache-busting query flood against a municipal services platform in the Middle East, peaking at 394.8 million requests per second. Mitigated in 0.141 seconds.
memcached amplification against a education platform platform in Iberia, peaking at 290.8 Gbps. Mitigated in 0.519 seconds.
HTTP POST flood against a crypto exchange platform in Western Europe, peaking at 776.9 million requests per second. Mitigated in 0.237 seconds.
Here is a question we could not answer well a year ago: what really happens with choosing between DNS delegation and CNAME onboarding? We can answer it now.What we changedWe moved the…
HTTP POST flood against a sports betting platform in Latin America, peaking at 638.8 million requests per second. Mitigated in 0.934 seconds.
Here is a question we could not answer well a year ago: what really happens with why our logs are append-only and how we query them? We can answer it now.Observability firstEvery…