The Observability Money Heist - a Fintech's Tale of...
summary
Fintech reduces observability and infra spend by replatforming concrete components and automating responses: swaps Prometheus/Thanos for VictoriaMetrics and cuts metric cluster cost from ~$60k to ~$15k (saving ~$40–45k), replaces Fluentd with Fluent Bit shippers plus Parquet-on-S3 retrieval to avoid full ingestion, abandons 100% tracing that cost ~$3M/year from HyperTrace, and automates common fixes with Shoreline runbooks and Zenduty routing, cutting alerts from ~11,000 to ~3,000 and lowering time-to-isolate by ~20%.
AI-written summary · about
chapters (12)
- 0:00Welcome
- 1:24Cost optimization matters
- 2:30Observability costs at scale
- 4:39Razorpay’s scale
- 5:42VictoriaMetrics savings
- 7:51VictoriaMetrics pricing
- 9:15Log shipping overhaul
- 10:18Tracing cost disaster
- 11:04eBPF for network visibility
- 12:11Incident response inefficiencies
- 13:35Automating runbooks
- 15:59Takeaways and Q&A
related talks
16mShrinking the Observability Bill
41mResilient Observability at the Retail Edge: A Lightweight, Scalable, and...
08mObserving the Invisible: Real-Time Financial Risk Detection
46mObservability in digital transformation
40mFrom Debt to Wealth: Transforming Observability at Adobe
13mNext-Gen Observability: reduce MTTR
47mCost optimization
30mIs Building an Observability Platform with Open Source the Right Choice?