Adopting Open Telemetry Collector @ eBay - Swapping Engines Mid Flight
summary
eBay describes replacing Metricbeat with OpenTelemetry Collector across Kubernetes, migrating 1.25 million endpoints that produce ~32 million samples/sec and 1.5 billion active series. They move agents from node DaemonSets to sharded cluster-level stateful sets, achieve 90% capacity savings, and implement a custom file reload receiver to mimic dynamic config reloading. The team validates scrape parity with side‑by‑side runs, files PRs for sanitizer and exemplar gaps, and automates regressions before rolling upgrades.
AI-written summary · about
chapters (4)
related talks
49mAdding OpenTelemetry to Production Apps Lessons Learned
37mHow To Adopt OpenTelemetry in an Enterprise Where Incumbent Vendor Tools Reign Supre...
25mOpenTelemetry Collector Deployment Patterns
16mUsing the OpenTelemetry Collector to Empower End Users
26mScaling OpenTelemetry for Modern Workloads: From Mobile To Ma...
29mFrom Statsd To OpenTelemetry: Atlassian's Metrics Platform Migration…
27mIngesting 6.5 Tb of Telemetry Data Daily Through Open Telemetry Protocol and Collectors
34mPanel: OpenTelemetry: Bridging Platform and Enablement