openobserve

Why we included this project

OpenObserve is a self-hosted observability platform that brings logs, metrics, and traces together in one place without the per-seat pricing of Datadog or Splunk. The storage design is what keeps the price down: columnar Parquet files on S3 cut storage costs by up to 140x, so you can keep months of history instead of trimming retention to fit a budget. If you build on LLMs, it also captures prompt and response traffic, so you can watch latency, token usage, and error rates on your AI features alongside the rest of your stack. Everything runs from a single binary, which a small team can stand up and operate without much fuss, and it ingests from OpenTelemetry, LangChain, or LangFuse instrumentation you may already have. Teams on Prometheus or OpenTelemetry can point those sources at it directly, so it slots into an established stack rather than demanding a rewrite.