JEP 307: Parallel Full GC for G1 에 대해서 - JDK 10에서 변경사항

원문 : http://openjdk.java.net/jeps/307

•  요약
• G1 GC worst-case에서 향상 목적
• JDK9부터 기본 GC가 G1으로 변경되었는데, concurrent GC에서 full GC가 발생 할 수 있음. 
• G1 GC의 알고리즘 mark-sweep-compact를 병렬로 수행하도록 변경되었으며 -XX:ParallelGCThreads 옵션을 사용해서 스레드 수 조정 가능

JEP 307: Parallel Full GC for G1

Owner	Stefan Johansson
Type	Feature
Scope	Implementation
Status	Closed / Delivered
Release	10
Component	hotspot / gc
Discussion	hotspot dash gc dash dev at openjdk dot java dot net
Effort	M
Duration	M
Reviewed by	Mikael Vidstedt
Endorsed by	Mikael Vidstedt
Created	2017/01/17 11:40
Updated	2018/03/29 07:39
Issue	8172890

Summary

Improve G1 worst-case latencies by making the full GC parallel.

Non-Goals

Match the performance of the parallel collector's full GC for all use cases.

Motivation

The G1 garbage collector was made the default in JDK 9. The previous default, the parallel collector, has a parallel full GC. To minimize the impact for users experiencing full GCs, the G1 full GC should be made parallel as well.

Description

The G1 garbage collector is designed to avoid full collections, but when the concurrent collections can't reclaim memory fast enough a fall back full GC will occur. The current implementation of the full GC for G1 uses a single threaded mark-sweep-compact algorithm. We intend to parallelize the mark-sweep-compact algorithm and use the same number of threads as the Young and Mixed collections do. The number of threads can be controlled by the -XX:ParallelGCThreads option, but this will also affect the number of threads used for Young and Mixed collections.

Testing

• Full GC time analysis to ensure that the full GC times have improved. Looking at benchmark scores will probably not be good enough since G1 is designed to avoid full GCs.
• Runtime analysis using VTune or Solaris Studio Performance Analyzer to find unnecessary bottlenecks.

Risks and Assumptions

• The work is based on the assumption that nothing in the fundamental design of G1 prevents a parallel full GC.
• The fact that G1 uses regions will most likely lead to more wasted space after a parallel full GC than for a single threaded one.