Datadog rebuilt its Git serving infrastructure with 'gitretriever' to handle a 20x increase in CI traffic. The new system uses independent pods with 'mirrors' to sync with GitHub and 'relays' to serve code to CI workloads. This architecture avoids the replication overhead of previous systems and reduces CPU load by distributing Git pulls across branches and optimizing packfile generation. Key principles include parallel fetching for changed references, serving local copies directly without conse
All Datadog trails