Memory Efficient Java

on YouTube ↗38m

" summary

Claims that roughly 60% of Java apps are memory-inefficient, shows concrete diagnostics and fixes: uses GC logs and profilers (VisualVM, Flight Recorder) to find hot allocation sites, measures allocation rate (examples: 6.5 GB/s and 4 GB/s cases), explains TLABs and escape analysis, and rewrites hot loops to eliminate frequent array/object allocations. Results are dramatic: reducing allocation from 1.8 GB/s gives throughput jump from 400k TPS to 25M TPS, a websocket case halves churn and boosts throughput, and a loop fix yields ~2x runtime improvement.

AI-written summary · about

" chapters (10)

1. 0:00Intro
2. 2:12Demo
3. 4:15What is your performance trouble spot?
4. 11:57War stories
5. 13:19Allocation site
6. 16:39Java heap
7. 20:12Problems
8. 22:22Pause vs. occupancy
9. 23:56Reducing allocations
10. 25:21Demo

" related talks

• 21mMemory Efficient JavaKirk Pepperdine · GOTO Chicago 2020
• 39mMemory Efficient JavaKirk Pepperdine · Devnexus 2020
• 51mA Healthy diet for your Java applicationMargarita Nedzelska · Voxxed Days Luxembourg 2023
• 56mJava Performance AnalysisMonica Beckwith, Kirk Pepperdine · Devnexus 2020
• 40mSave Java memoryJan Kotek · JavaDay Kyiv 2015
• 45mHow difficult can it be to write efficient code?Roberto Cortez · Voxxed Days Zürich 2025
• 02h 41mSolving Memory Leaks in the JVMKirk Pepperdine · Devoxx Belgium 2019
• 35mJava *is* Memory EfficientNicolai Parlog, Ron Pressler · JavaOne 2026