The full chain load test in AliExpress Russia
summary
AliExpress Russia runs a full-chain load test that models peak QPS using a formula with recent campaign peaks and a campaign adjustment coefficient, splits read/write flows across Russia and a Germany backup DC, and prepares realistic requests by extracting parameters from production logs into shadow tables. A cloud pipeline uploads test files to Alibaba OS, distributes them to executors, and supports one-button start/stop and manual ramp adjustments; the full one-hour data prep pipeline completes in practice. Tests reveal problems like a 5x downstream spike after endpoint refactor and missing traffic tag propagation across threads, both fixed.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:14Important terms
- 2:07About AliExpress Russia
- 2:57Define consumer and service
- 4:13Analyze traffic model
- 6:55Read infoction
- 8:19Data preparing
- 9:21Pipeline overview
- 10:01File upload
- 11:07Load test
- 12:46Low test
- 13:39Low test monitor
- 15:04Dashboard
- 15:51Classic problems
related talks
23mAI-Driven Load Test Analysis: From Routine Toward Action
37mExploratory Load-Testing with AI Tooling - it's Time to Ramp Up!
31mLoad Testing Kubernetes: How to Optimize Your Cluster Resource Allocation in Production [I]
46mChaos Experiments - Datacenter Stress Testing
44mSofts You Need to Run the Perfect Load Test
17mDeep dive on Azure Load Testing in CI/CD
01h 08mPerformance testing in the cloud
48mTesting The Cloud