Performance Testing with a "Big Data Tool"
summary
A Splunk-based system replaces slow PDF reports and ingests millions of timing logs to produce dynamic dashboards that index 100 million lines in 10 seconds, let teams attach an environment in 20 seconds, and share problems via a unique interactive URL. Overlapping JSON TPS tracing with JVM garbage-collection logs exposes root causes; a clearing-house KPI of 60 position updates/sec (≈17 ms) moves from a 99 ms average down to 14 ms. Rollout includes one-click uploads, three-month retention, and real production fixes at a client clearing $1 trillion daily.
AI-written summary · about
chapters (8)
- 0:00Performance testing intro
- 4:45Splunk for testing
- 8:41Java tracing issues
- 19:48Banking performance problem
- 24:09Overlapping data sets
- 31:52Visualizing test results
- 33:17Big data tool requirements
- 35:50Q&A on Splunk
related talks
40mPerformance Testing and Big Data
42mAutomation Testing in a Real Big Data Project
56mBetting on Performance: A note on Hypothesis-driven Performance Testing
33mContainerized Time Travel: Replicating Production Performance
07mSmart Load Testing
21mCrunching Millions of Requests for Breakfast: A Tale of Performance Testing Prowess
49mPerformance Testing
45mPerformance Testing and Tuning of Web Applications