How we improved availability through iterative simplification

This post details specific instances of improving availability through iterative simplification. It describes optimizing a SQL query for GitHub's Command Palette by reordering joins, resulting in an 80-90% performance improvement, and further optimizing a filtering query by batching, yielding another 20-80% improvement. It also covers simplifying a redirect endpoint by removing unnecessary access checks, controlled via a Flipper feature flag, which improved P75/P99 request latency and reduced max latency. The post highlights the use of Datadog, Splunk, Scientist, and Flipper as key tools in this process.