Home > Java > javaTutorial > Java Virtual Machine Learning - Memory Tuning

Java Virtual Machine Learning - Memory Tuning

黄舟
Release: 2017-02-17 10:32:00
Original
1623 people have browsed it

JVM tuning is mainly for memory management tuning, including controlling the size of each generation and GC strategy. Since GC will suspend application threads when it starts garbage collection, which seriously affects performance, the purpose of tuning is to minimize the pause time of application threads caused by GC and reduce the Full GC times.

Generation size tuning

The most critical parameters: -Xms, -Xmx, -Xmn, -XX: SurvivorRatio, -XX:MaxTenuringThreshold, -XX:PermSize, -XX:MaxPermSize

-Xms, -Xmx are usually set to the same value to avoid continuously expanding the JVM memory during runtime. This The value determines the maximum memory that the JVM heap can use.

-Xmn determines the size of the new generation space. The ratio of the three areas of the new generation Eden, S0, and S1 can be controlled by -XX:SurvivorRatio (if the value is 4, it means: Eden :S0:S1 = 4:3:3 )

##-XX:MaxTenuringThreshold controls how many times the object enters the old age after minor GC. Times, this parameter is only valid in Serial serial GC.

-XX:PermSize, -XX:MaxPermSize are used to control the size of the method area, usually set to the same value.

1. Avoid setting the size of the new generation too small

When the new generation setting is too small, two obvious phenomena will occur. One is that the number of minor GCs is frequent, and the other is that minor GC objects may directly enter the old generation. When there is insufficient memory in the old generation, Full GC will be triggered.

2. Avoid setting the new generation too large

If the young generation is set too large, it will bring about two problems: first, the old generation becomes smaller, which may lead to frequent execution of Full GC; second, the time for minor GC to execute recycling increases significantly.

3. Avoid the Survivor area being too large or too small

-XX: The greater the value of the SurvivorRatio parameter, it means that the Eden area becomes larger and the number of minor GCs will decrease, but the two Survivor areas become smaller. If Objects that exceed the Survivor area memory size are in minor If it is still not recycled after GC, it will directly enter the old generation,

##-XX: The SurvivorRatio parameter value is set too small, which means that the Eden area becomes smaller, the number of minor GC triggers will increase, the Survivor area becomes larger, which means that more can be stored in the minor The surviving objects after GC are prevented from entering the old age.

4. Reasonably set the survival period of the object in the new generation

The value of the new generation survival cycle determines how many times the new generation object enters the old generation after undergoing Minor GC. Therefore, this value should be tuned according to your own application. This value in the Jvm parameter corresponds to -XX:MaxTenuringThreshold, and the default value is 15 times.

GC tuning

1.GC matching

See another garbage collector

2. View JVM parameters

Command line tool Visual tool

3.GC log parameters

-XX:+PrintGCDetails

##-XX:+PrintGCTimeStamps

##-Xloggc:c:\java\gc.log


The above is the content of Java virtual machine learning - memory tuning. For more related content, please pay attention to the PHP Chinese website (www.php.cn)!


source:php.cn
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template