JVM Optimization

Understand Your Application's Behavior

Profile Your Application

Use profiling tools to understand memory usage patterns and identify memory leaks. VisualVM can be used for the same

        

https://github.com/manaspratimdas/memory-analysis/tree/master/myappsmem/src/main/java/myappsmem/heapexhaustion/ml

 

Monitor GC Logs

Regularly analyze GC logs to understand the frequency and duration of GC. While running the application we can enable GC logging with below JVM argument as follows 

-Xlog:gc*:file=gclog/gc_%t_%p.log

Choose the Right Garbage Collector

•    Serial GC: -XX:+UseSerialGC: Suitable for small applications with low memory requirements.
•    Parallel GC: -XX:+UseParallelGC : Good for applications with high throughput requirements.
•    CMS GC: -XX:+UseConcMarkSweepGC : Suitable for applications requiring low pause times.
•    G1 GC: -XX:+UseG1GC : A balanced option for applications with large heaps and requiring predictable pause

      Illustration: Serial GC vs G1 GC

https://github.com/manaspratimdas/memory-analysis/blob/master/myappsmem/src/main/java/myappsmem/optimization/MyAppMemAnalyzer.java

Run the program with JVM arguments 

• -Xlog:gc*:file=gclog/gc_%t_%p.log -XX:+UseSerialGC
• -Xlog:gc*:file=gclog/gc_%t_%p.log -XX:+UseG1GC

Tune Heap Size

Set Initial and Maximum Heap Size

Use `-Xms` and `-Xmx` to set the initial and maximum heap size. It's often recommended to set them to the same value to avoid resizing during runtime

• -Xlog:gc*:file=gclog/gc_%t_%p.log (Default) [ it took xms as 254 and xmx as 4048]
• -Xms16m -Xmx512m -Xlog:gc*:file=gclog/gc_%t_%p.log
• -Xms64m -Xmx64m -Xlog:gc*:file=gclog/gc_%t_%p.log

https://github.com/manaspratimdas/memory-analysis/blob/master/myappsmem/src/main/java/myappsmem/optimization/MyAppMemAnalyzer.java

Adjust Young Generation Size

Use `-XX:NewSize` and `-XX:MaxNewSize` to tune the size of the young generation. A larger young generation can reduce the frequency of minor GC

• -XX:NewSize=128m -XX:MaxNewSize=128m: No GCs occurred, indicating that the memory allocation was sufficient to avoid GC events.
• -XX:NewSize=64m -XX:MaxNewSize=64m: One GC event occurred, suggesting that the memory allocation was almost sufficient but required one cleanup.
• -XX:NewSize=16m -XX:MaxNewSize=16m: Seven GC events occurred, indicating that the memory allocation was insufficient, leading to frequent GCs.

https://github.com/manaspratimdas/memory-analysis/blob/master/myappsmem/src/main/java/myappsmem/optimization/YoungGenerationTest.java

When the number of garbage collections (GC) increases, it can significantly impact the performance of your application

• GC Pause Time: Frequent GCs cause more pauses, degrading application responsiveness and throughput.
• CPU Usage: Higher GC frequency increases CPU usage, as more time is spent on memory management.
• Latency: More frequent GCs lead to higher latency, affecting real-time performance.
• Memory Fragmentation: Frequent GCs can cause memory fragmentation, slowing down memory