Should I use OTel (collectors), or is Prometheus good enough?
summary
Answer is: it depends. OpenTelemetry Collector can use a Prometheus receiver to scrape metrics, tail logs and export to Tempo, Thanos, DataDog or any backend, but it is not a 100% replacement for Prometheus — no local PromQL, no built-in alerting or remote_read and sharding must be handled. At scale (tens of millions of users, hundreds of Prometheus instances across on‑prem and k8s) a hybrid approach often wins. Solo.io’s pattern of a collector gateway design (workload collectors + management‑cluster gateway) centralizes processing and works for multi‑tenant exports, but check vendor ingestion costs and failure modes first.
AI-written summary · about
chapters (9)
- 0:00Introduction and overview
- 5:28OpenTelemetry 101
- 14:48State of observability before OTel
- 16:57Promise of OTel collectors
- 22:20Scenario one: Cloud infrastructure
- 26:56Scenario two: Vendor agents
- 28:41Scenario three: Vendor solutions
- 31:58Key takeaways and tradeoffs
- 32:41Q&A
related talks
37mShould I Use OTel (collectors), or Is Prometheus Good Enough?
24mLeveraging OpenTelemetry for Your Prometheus Pipeline
10mUpstream-First, High Scale Prometheus Ecosystem
27mThe State of Prometheus and OpenTelemetry Interoperability
22mBuilding Prometheus Metrics Support in OpenTelemetry
25mDebugging the Pipeline: Observability for Your Observability with OTel + Prometheus
29mWho Watches the Watchers? From Closed Observability To...
27mObservability with Prometheus and beyond