Applied Software Engineering Case Diary: Field Lessons in Problem Solving
summary
Field cases show concrete fixes and surprising scale: adjust IBM Samura JVM minor-heap sizing to stop frequent minor/major GCs; fix Apache Spark on PPC by making a one-line serializer change to correct endianness and save $70 million at stake; resolve thousands-of-threads hangs by forcing volatile semantics when JIT over-optimizes; explain why RSS memory stays high across 200+ AWS installs; diagnose rare R1 corruption from cache store reordering and eliminate it with a sync. reproducible small test cases
AI-written summary · about
chapters (8)
- 0:00Introduction to field lessons
- 1:07Business impact vs problem complexity
- 2:53JVM performance degradation
- 5:22Apache Spark on PPC servers
- 9:17JUC hang in e-commerce
- 21:51RSS memory retention issue
- 28:27R1 register corruption mystery
- 35:52Key takeaways and conclusion
related talks
40mProblem Solving Recipes Learned from Supporting Spark
40mDebugging JVM performance issues
47mHunting issues in a Java app with 31G heap and 500G off-heap: an Apache Ignite case study
41mOptimising Apache Spark and SQL for Improved Performance
38mFinding a needle in the haystack - Debugging performance issues
36mHow to make your daily routine, code and database more efficient
51mThe Diabolical Developer's Guide to Perfomance Tuning
40mScaling with Apache Spark (or a lesson in unintended consequences)