HTTP resilience tradeoffs in practice: retry vs Retry-After vs hedging
I ran 3 controlled chaos scenarios comparing retry-only, Retry-After-aware retry, and hedging.
One run improved success with retries but significantly worsened p95/p99 under a tight timeout budget. Another showed that honoring Retry-After turned a 40% 429-heavy error profile into stable completion. More in the post. #programming #dev #js #technology