JVM (XX) -- performance monitoring and tuning (I) -- Overview

1. Interview questions of Dachang

1. JVM What does performance tuning do ？
2. I did JVM Memory optimization ？
3. from SQL、JVM、 framework 、 Four aspects of database optimization ideas
4. JVM Compiler optimization
5. JVM What diagnostic tuning tools have been used
6. jvm How to tune 、 Heap memory 、 How much stack space is appropriate
7. JVM What are the related analysis tools used ？ What are the specific performance tuning steps
8. jvm How to tune , How to set parameters
9. How to understand the memory leak problem , What can cause memory leaks ？ How to solve
10. Why do hundreds of thousands of concurrent second killing systems happen frequently every second GC
11. How to optimize the daily average million level trading system JVM
12. On line production system OOM How to monitor, locate and solve
13. How high concurrency system is based on G1 Recycler optimizes performance

2. The background that

2.1 Problems in the production environment

1. How to deal with memory overflow in the production environment
2. How much memory should the production environment allocate to the server
3. How to optimize the performance of garbage collection period
4. Production environment CPU How to deal with high load
5. How many threads should the production environment allocate to the application ？
6. No addition log, How to determine whether a line of code has been executed by a request
7. No addition log, How to view the return value of a method in real time

2.2 Why tuning

1. Avoid OOM
2. solve OOM
3. Reduce Full GC Frequency of occurrence

2.3 Different stages of consideration

1. Before going online
2. Project operation phase
3. Appear online OOM

3. Tuning Overview

3.1 The basis of monitoring

1. Run log
2. Exception stack
3. GC journal
4. Thread snapshot
5. Heap dump snapshot

3.2 The general direction of tuning

1. Write code properly
2. Make full and reasonable use of hardware resources
3. Reasonably carry out JVM tuning

4. Performance optimization steps

4.1 Find the problem ： Performance monitoring

1. GC frequent
2. CPU load Too high
3. OOM
4. Memory leak
5. Deadlock
6. The program response time is too long

4.2 Troubleshoot problems ： Performance analysis

1. Print GC journal , adopt GCviewer To analyze log information
2. Flexible use of command line tools ,jstack,jmap,jinfo etc.
3. dump Get out of the pile , Use memory analysis tools to analyze files
4. Use Ali arthas perhaps jconsole、jvisualVM To implement the view JVM state
5. jstack View stack information

4.3 solve the problem ： performance tuning

1. Increase memory appropriately , Choose garbage collector according to business in Beijing
2. Optimize the code , Control memory usage
3. Add machines , Disperse node pressure
4. Reasonably set the number of threads in the thread pool
5. Use middleware to improve program efficiency , Such as caching , Message queuing, etc

5. Performance evaluation 、 Test indicators

1. Pause time （ Or response time ）
   

2. throughput

   1. The brightness of the workload completed in unit time
   2. stay GC in ： The percentage of time spent running user code in total runtime （ Total operation time = The running time of the program + Time of memory recovery ） Figure two 1 - 1/(1 + n).-XX:GCTimeRatio = n

3. Concurrency number ： At the same time , The number of requests that actually interact with the server

4. Memory footprint ：java The amount of memory the heap occupies