Row-watching performance for different roles

The Technical Department generally has the following common roles: developers, testers, architects (system administrators, DBAs), and O & M personnel.

 

1. Black box testing

Black box test operation application interface-Data Request sent over the network-server receiving and processing-get relevant data on DB server-return data after processing in the previous section-application interface receives data to respond to the next step.

The black box test only cares about the single-step response time of the application. The performance depends on the application time, that is, the sum of the round-trip time after the data stream passes through the server/server cluster over the network.

 

2. Development Perspective

1) architecture Rationality

2) rationality of Database Design

3) Code

4) memory usage in the system

5) thread usage in the system

6) Is there any vicious or irrational competition in system resources'

7) as a developer, the developer only pays attention to the Code Implementation of the function and seldom pays attention to the database design. Is the framework design reasonable, whether the memory usage in the system is reasonable, whether the thread usage in the system is reasonable, and whether there is unreasonable competition for system resources. Usually people think that this is a problem for architects, and small companies will not pay attention to these issues.

 

3. System Administrator

1) hardware resource utilization

2) JVM

3) dB

4) which hardware can improve system performance?

5) Can the system support services?

6) scalability, compatibility, maximum capacity, and possible bottlenecks

7) as O & M personnel, they usually pay attention to whether all servers in the system are running normally. They generally pay attention to the hardware resource utilization of these servers (databases, middleware, and other servers), such as whether the memory has available space, whether the CPU exceeds 70%, whether the network is smooth, and whether there is a bottleneck in I/O. Can these servers and configurations support stable and trouble-free operation of the system for several months or even years? In addition, we also consider increasing throughput requirements as the company's business grows, you can increase the overall throughput by adding servers.

 

4. Performance Testing

1) server hardware performance

2) set performance goals based on requirements and historical data

3) establish a performance pass Model

4) Performance Analysis on the Development Framework and hardware framework

5) benchmark tests for development and release versions

6) production environment configuration and Optimization

7) Develop test cases for Performance Testing

8) perform software performance acceptance and stability tests

9) Design Performance testing scenarios

10) coordinate with various departments

11) Specific Performance Analysis

 

Note:

1. What does throughput mean?

[Data processing volume per second]Throughput is the number of transactions completed by the system within a given period of time. That is to say, the higher the system throughput, the more users or system requests the system has completed per unit time (measured in bits, byte, group, etc.), and the system resources are fully utilized.

 

Row-watching performance for different roles