The key evaluation index of system performance test

Usually, the common indicators to measure the performance of a software system are:

1. Response time (Response time)
Response time is the time it takes for users to feel the software system serves them. For a website system, the response time starts when a page is clicked and ends when the page is completely displayed in the browser. This time interval seems simple, but in fact, during this response time, the software system undergoes a series of processing work behind the scenes, running through the entire system node. Depending on the "jurisdiction", the response time can be subdivided into:

(1) The server-side response time, which refers to the time for the server to complete the execution of the transaction request, excluding the client-to-server response (request and time spent on the network) communication time), this server-side response time can measure the processing power of the server.
(2) Network response time, which is the time it takes for network hardware to transmit transaction requests and transaction results.
(3) Client response time, which is the time it takes the client to construct the request and display the transaction result. For ordinary thin client web applications, this time is very short and can usually be ignored; but for fat clients For client-side web applications, such as Java applet and AJAX, since a large amount of logic processing is embedded in the client, it may take a long time and become the bottleneck of the system. This is a point to be paid attention to.
Then the response time felt by the client is actually equal to the client response time + server response time + network response time. The purpose of subdivision is to facilitate locating on which node the performance bottleneck occurs (what is the performance bottleneck, which will be introduced in the next section).

2. Throughput
Throughput is a common software performance indicator. For a software system, what is "swallowed" is the request, and what is "spit" is the result, and the throughput reflects the "food intake" of the software system, that is, the system's Processing capacity, specifically, refers to how many transactions/requests/units of data the software system can process per unit of time. However, its definition is flexible and has different interpretations in different scenarios. For example, the throughput of the database refers to the number of executions of different SQL statements per unit time; the throughput of the network refers to the transmission on the network per unit time. data traffic. The amount of throughput is determined by load (such as the number of users) or behavior. For example, downloading files requires higher network throughput than browsing the web.

3. Common resources for resource utilization
include: CPU usage, memory usage, disk I/O, and network I/O.

4. Hits HPS (Hits per second) Hits
are a useful indicator to measure the processing power of a Web Server. It needs to be clear that the number of clicks is not the number of mouse clicks that we usually understand, but is calculated according to how many http requests the client initiates to the Web Server. One mouse may trigger multiple http requests, which needs to be combined with the specific Web system. implemented to calculate.

5. TPS - TPS: Transaction per second, the number of transactions or transactions that the system can process per second. It is an important indicator to measure the processing capability of the system.

6. The number of concurrent users (Concurrent users)
The number of concurrent users is used to measure the concurrent capacity and synchronization coordination ability of the server. On the client side refers to a group of users performing an operation at the same time. The number of concurrency reflects the concurrent processing capability of the software system. Unlike throughput, it mostly occupies operating system resources such as sockets and handles.

In addition, the performance indicators of the software system and the system recovery time are measured. In fact, all the user's requirements on resources and time can be regarded as performance indicators and can be used as the measurement of the software system. The performance test is to verify whether these performance indicators are satisfied.