New approach to metrics observability for the cloud era
summary
Claims a production system built on a custom S3-native store and serverless query Lambdas that retains full-fidelity metrics (no downsampling) and serves a demo workload of ~15M unique time series per hour (~1.5B samples/day). Reports ~10x lower compute costs and ~24x lower storage costs versus SSD/EBS approaches, with a demo cost of ~$4M/year versus ~$10M on Datadog. Provides dashboard-less alert debugging via deterministic dimensionality-impact scans that surface root causes (e.g., DB connection 500s) in minutes.
AI-written summary · about
chapters (10)
- 0:00Introduction to Oodle and Speaker Background
- 1:15Current Challenges in Metrics Observability
- 1:53Typical Metrics Observability Stack
- 3:09In-Depth Look at Observability Challenges
- 4:14Addressing Observability Challenges
- 10:49Requirements for an Ideal Solution
- 13:05Why Now is the Right Time to Rethink Observability
- 14:37Oodle's Innovative Approach and Demo
- 19:16Cost Efficiency and Debugging Speed
- 22:10Summary and Conclusion
related talks
15mThree key considerations for deploying best-of-breed observability
52mReal-time-first metrics
33mNetdata: Open Source, Distributed Observability Pipeline – Journey and Challenges
34mObservability in Serverless Application
40mLarge-Scale Logging Made Easy
13mNext-Gen Observability: reduce MTTR
32mPowering cloud modernization w/log observability in context & at scale
36mFinOps & Observability: technical implementation