Lightning Talk: Rate Limiting With Quotas Is Hard - So We Didn't!
Lightningat EnvoyCon North America 2025 · on Monday, 2025-11-10· at 11:45· Building B | Level 4 | B406b-407·
summary
Tesco runs Envoy across multiple regions for ~800 APIs and ~100,000 store checkouts and abandons RLQS quotas because per-envoy allocations lag during rapid traffic shifts. A custom protocol sends a single percentage-based approach to every Envoy, implemented via a local sidecar translator in the same pod so decisions stay low-latency. Central processing and data volumes drop by orders of magnitude, global rate limits stay accurate during redeploys and drains, and the rollout shows zero false positives while traffic rebalances.
AI-written summary · about
related talks
10mLightning Talk: Closing the Gap: Fair North–South Bandwidth Management for Tenants
27mAdventures in Rate Limiting: Spotify’s Journey Writing a Scalable Envo...
08mLightning Talk: It’s 10pm, do you know what your proxy is doing? Understanding...
29mSpanning the Globe with Envoy at Stripe
11mLightning Talk: Building a Scalable Multi-Protocol API Gateway with Envoy
25mBuilding Low Latency Topologies with Envoy
21mMaking Envoy Resilient to Sudden Increases in Load
06mLightning Talk: How a Global Logistics Company Migrated Clouds with Zero Downtime