52 Minutes a Year: What 99.99% Availability Actually Costs
Everyone puts 99.9% in their SLA. Few know what it actually costs to hit it — and almost nobody distinguishes between a service that's 'up' and one that's actually reliable.
7 min read
5 articles with this tag.
Everyone puts 99.9% in their SLA. Few know what it actually costs to hit it — and almost nobody distinguishes between a service that's 'up' and one that's actually reliable.
7 min read
Everyone talks about scaling. Few talk about the specific moment their system broke, why it broke, and the non-obvious fix that actually worked. Here's ours.
7 min read
Fixed and sliding windows argue about counting; token and leaky buckets argue about bursts. Here’s the contract each one enforces — and why production systems layer both.
6 min read
Fixed window rate limiting looks simple and behaves well on dashboards — until traffic piles up at window boundaries. Here’s why the contract is met while the system still overloads.
3 min read
Sliding windows answer the right question — how many requests in the last N seconds — but log-based precision doesn’t scale, and bucketed counters still miss microbursts.
4 min read