LodeHQ
AI news for people who shipFree, by email

DevOps · datadoghq.com · Wednesday, 19 August 2026

Datadog's gitretriever serves 100M+ CI fetches a week at ~40ms

Datadog's answer to CI Git slow-downs: a purpose-built read mirror that survived 20x traffic growth with ~40ms median latency, handling 100M+ fetches weekly. By offloading reads from its GitLab backend, it slashed fetch-serving CPU three-to-fourfold and tamed the monorepo outages AI coding agents were triggering.

Read the source: datadoghq.com ↗In the 19 August issue →