(1)
Confirm the problem
Determine whether the system is faulty Based on the test results. If yes, determine what the problem is. And correctly locate the problem, focusing on identifying system bottlenecks. Generally, start with the following points:
· Application code: Generally, many program performance problems are written. Therefore, you should first check the code for the module that discovers the bottleneck.
· Database Configuration: Database Configuration often causes slow operation of the entire system.OracleFor large databases, DBA (Database Administrator) must make correct parameter adjustments before production.
· Operating System Configuration: unreasonable operating system configuration may also cause system bottlenecks.
· Hardware settings: disk speed, memory size, and so on are all likely to cause bottlenecks, so these are the focus of analysis.
· Network: network conflicts and network latency may occur when the network load is too heavy.
At the same time. We also need to investigate the usage of the system,
· Have you heard many complaints from users?
· Does the response time of some operations become longer with the longer time?
· Is the CPU usage very low? Is the I/O usage very high?
· Is the performance stable during use?
The system performance problem is not obvious. You need to perform a careful search to locate the problem correctly.
(2) determine the cause
After confirming that the system has a problem, analyze it carefully to determine the cause of the problem. The reason for determination depends largely on personal experience. The knowledge involved includes operating systems, databases, networks, and programs.DevelopmentAnd so on.
As with determining performance issues, identifying causes still requires extensive information collection. Perform the following analysis:
· What is the impact of the problem: response time, throughput, or other problems?
· Are most users or a few users having problems? If there are a few users, what are the differences between these users and other users?
· Is the result of system resource monitoring normal: Has the CPU usage reached its limit? What is the I/O condition?
· Is the problem concentrated in a certain type of module?
· Is there a problem with the client or server?
· Is the system hardware configuration sufficient?
· Does the actual load exceed the system's load capacity?
· Is the system not optimized?
Through these analyses and some specific system-related problems, you can have a deeper understanding of the System Bottleneck and analyze the causes.
(3)
Determine adjustment targets and Solutions
After analyzing the cause of the problem,TesterAnd other system adjustment personnel should first determine the adjustment target, and then design a solution, determine the adjustment target, the main role is to determine when to stop the adjustment system, otherwise the work will never be banned.
Each system has different features, so the adjustment targets may be different. For example, the following are the goals of system adjustment:
· Improve system throughput.
· Shorten the response time.
· Better support for concurrency.
The main basis for designing a solution is the adjustment goals with clear solutions and objectives. You can perform subsequent work.
(4)
Test Solution
After implementing the solution, we need to test the solution. You can use the previousTest CasesTo verify whether the system solves the performance problem. The testing solution should be carried out in a simulated environment as far as possible, because it may cause damage in the production environment unless the test risk is fully estimated and a sufficient remedy is prepared.
(5)
Analyze adjustment results
The last step of performance adjustment is to analyze the adjustment results. If the problem is not solved, repeat the previous work. During the process of testing the system adjustment scheme, you should analyze the work you do frequently, for example, failing to locate the problem accurately or the adjustment scheme is incorrect,
The expected goals may not be met. Early discovery of these errors can bring the work back to the correct track.
The following problems are mainly considered in the analysis and adjustment results:
· Has the system adjustment reached or exceeded the intended target?
· Has the overall performance of the system been improved, or does it sacrifice some performance to solve the problem?
· Can the adjustment be completed?
Finally, if you reach the expected goal, the adjustment will basically end.