Real World SLOs and SLIs: A Deep Dive
summary
Walks through concrete SLI/SLO recipes used at New Relic: measure data ingest by counting 200 responses and message-bus latency by comparing ingest timestamps to availability, treat NRDB query latency as a 99th percentile SLI, and set per-component SLOs (services ~99.9%, infra ~4 nines). Hard-sharded systems get per-shard SLIs, container scheduler health uses link-saturation and job-state transitions (Mesos/Marathon) as signals, violations become alert thresholds and feed a synthetic end-to-end script that validates the overall customer-facing SLO.
AI-written summary · about