Introduction to test performance (performance indicators, performance test methods, performance testing tools ...)

First, the concept of performance tests

  • Performance: Transaction, certain characteristics of the evaluation value items
  • Performance Test: to test the system performance is simulated variety of normal, abnormal and peak load conditions by testing tool

Second, the performance test indicators

Performance indicators divided into two areas:

  • System metrics (user scenarios and requirements related indicators)
  • Resource indicators (and hardware resource consumption related indicators)

1. Response time

Initiation request received from the request response time

Response time = time + response network application response time = (N1 + N2 + N3 + N4) + (A1 + A2 + A3)

2. Concurrent

The number of user-initiated requests per unit of time

Concurrent users C, formula C = nL / T

The number of users accessing the system every day: n

L: Online users log in from time to quit

T: the user to use the system every day about how long

Peak C1, i.e. the maximum number of concurrent calculation formula C1 = C + ³√C

The optimum number of concurrent users: When the system load is equal to the optimum number of concurrent users, the highest overall efficiency of the system, no resources are wasted, users do not need to wait for
the maximum number of concurrent users: the system load continues, some users in treatment some users wait in the waiting time when their best

3. throughput, throughput

An important indicator of network performance

Throughput: amount of data transmitted over the network (process client requests)

Throughput: amount of data successfully transmitted to the network per unit time (which may be second / minute / hour / day), such as requests / sec, the number of pages / sec

4. affairs, TPS

Services: collection of an action or series of actions, such as user log on to a scene from exit on to a transaction

TPS: Transaction per second-- the most important performance indicators to measure server capacity to handle the number of transactions, the number of transactions per second by the

5. clicks, click-through rate

Hits: HTTP Web Server refers to the number of requests received

Hits: Hits Per Second, HTTP user per unit time to submit requests to the Web Server

Distinguish between mouse clicks: If a web page request, the page containing the three pictures, hits to the Web Server request: 1 + 3 = 4, and a click of the mouse you can access a Web page, only 1 hits

6. resource utilization

The main data processing computer and the computer instructions interpretation software: CPU

Memory: cpu bridge of communication with the computer running all programs are carried out in memory, the memory is divided into physical memory page swapping (Paging), SWAP memory (virtual memory)

  • Page swapping: When the physical memory that is the actual memory is full when the physical memory is not commonly used in the process of recall is stored in virtual memory to ease the pressure on the physical memory space, so when physical memory and virtual memory data exchange frequent time, this time we should focus on the case of memory performance
  • SWAP memory: virtual memory space allocated for the process, the hard disk space that is invoked as memory usage

Disk I / O: the amount of data per unit time through the disk. The main focus of the disk busy rate, if higher than 70%, the disk bottleneck

Network I / O: the amount of data per unit time over a network. When the device or network throughput is greater than a maximum transmission capacity of the link, i.e. the bandwidth, the network devices should consider upgrading or increasing bandwidth, Linux command netstate

Third, the performance test methods

Load testing, stress testing and performance testing, from testing methods and tools, the three are the same, however, stress tests with load testing belong to a subset of the performance test

1. Load Test

Under certain hardware and software environment, to determine the maximum number of users in meeting performance indicators able to withstand increasing loads through continuous. So its purpose is to obtain the maximum number of users. Generally not more than 80% cpu (28 principles), the maximum number of data users normally work

2. Stress Test

Under certain hardware and software environment, the server resources to the limit in the state by means of high load, test the system is stable in the long-running limit state. Including system metrics, server performance

Premise Fourth, the performance test

1. Necessity of performance testing - a key assessments

  • Authorities, regulatory review
  • Involving the safety of life and property
  • Large new system
  • Core System
  • Structure adjustment
  • Booms
  • Major bug fixes

2. testability

Performance indicator value can be quantified as

Fifth, performance testing tools

  • Jmeter
  • LoadRunner
  • Since the development of research tools

Guess you like

Origin www.cnblogs.com/xiaogongjin/p/11873143.html