Unlocking Cost Savings & New Possibilities: Your Guide to Prometheus Remote Write 2.0
summary
Remote Write 2.0 adds string interning plus native support for native histograms, exemplars, created_timestamps and per‑series metadata, and new headers to report partial write success; implementation is experimental but ready for adopters. Benchmarks show payload reductions often >50% (uploader claims up to 60% egress), serialization CPU down ~70% and allocation drops approaching 90% when converting classic histograms to native buckets. Upgrade requires a Prometheus feature flag and receiver support; client libraries and test tools need updates but the design stays stateless and simple to adopt.
AI-written summary · about
chapters (10)
- 0:00Introduction to Prometheus Remote Write
- 3:32History and evolution of remote write
- 11:13Requirements for Remote Write 2.0
- 15:10Development process of Remote Write 2.0
- 16:14Key improvements in Remote Write 2.0
- 18:22String interning for efficiency
- 20:09Benchmarking Remote Write 2.0
- 22:17Adopting Remote Write 2.0
- 24:01Summary and future plans
- 25:02Q&A session
related talks
42mPrometheus Intro and Deep Dive
28mIntegrating Long-Term Storage with Prometheus [A]
21mThe Future is Bright, the Future is Prometheus Remote Write
27mThe State of Prometheus and OpenTelemetry Interoperability
33mObservability Projects Updates
36mDeep Dive: Prometheus
36mPrometheus - Intro, Deep Dive, And Open Q+A
32mPrometheus Intro and Deep Dive