Riot Games: Evolution of Observability at the Gaming Company
summary
Riot replaces scattered homegrown collectors with Vector and cluster agents, injects telemetry via a dynamic admission webhook, and enforces sampling and allow‑lists to cut noise. The pipeline runs across 100+ Kubernetes clusters in 30+ regions, samples some logs at ~10%, and dynamically adjusts trace sampling by service. Results: logs drop from petabytes to ~350 TB/month, synthetics fall from 3M to 300k, custom ingestion drops ~90%, with ~30% vendor cost reduction and $5M projected cost avoidance.
AI-written summary · about
chapters (8)
- 0:00Intro
- 0:46What is Riot
- 1:52Technologies
- 5:06History
- 7:00Problems
- 10:59Tech Stack
- 18:27Migration
- 20:30Learnings
related talks
32mTaming the Cost of Telemetry: How Riot Games Reined In...
25mParity in Observability Across Different Platforms
13mNext-Gen Observability: reduce MTTR
41mResilient Observability at the Retail Edge: A Lightweight, Scalable, and...
43mFrom Zero to Billions of Traces: Distributed Tracing at Winamax
41mScaling Observability: From Chaos to Clarity with Open Source Tools
27mIngesting 6.5 Tb of Telemetry Data Daily Through Open Telemetry Protocol and Collectors
21mFrom Black Box to Glass Box: AI-Driven Observability for Modern APIs
