Analyze Website Health with Windows Performance Monitor Test

Test purpose and method

This is an API interface website, which can be logged in anonymously. It is deployed on IIS. The test is divided into three cases. The first case is that the interface executes a large number of queries, and it takes about 10 seconds to complete the query. The test results are shown in column 1 in the following chart. The second case is that the interface performs a null operation, and the test situation is shown in the second column in the chart. The third case is that the interface performs normal business, and the time to complete the normal business is less than 1 second. For the test situation, see the third column in the chart. The website has a stable user traffic at any time. The test is carried out on the server where the website is deployed. When testing the first case, change the interface code of the website to the code that executes a large query, and start the test after publishing, deploying, and starting the website. For the test method, see my other blog post Check the traffic situation of the iis website . When testing the second case, change the interface code of the website to empty code, republish, suspend the website, replace the folder content of the website with the newly released website folder content, restart the website, and then Test, the test process of the third case is similar to the second case.

Each test mainly records and analyzes the four indicators of the website's Anonymous Users/sec, Current Connections, Logon Attempts/sec, and Connection Attempts/sec, see lines 1, 2, 3, and 4 in the chart, and the currently tested indicators The change graph is black, for example, the first row shows the Anonymous Users/sec indicator, and the black graphs in the images in columns 1, 2, and 3 of the first row represent Anonymous Users/sec in the first, second, and third cases Changes in indicators. The test time for each test is 1 minute and 40 seconds.

Test Situation Analysis

According to the charts of the test records, some conclusions can be drawn as follows:

For the Anonymous Users/sec (anonymous users per second) indicator, the first case has a larger change range than the second and third cases, showing that there is often a large congestion of users per second.

For the Current Connections (current connections) indicator, the first case quickly climbed to a high value after the website was launched, and remained high, while the second and third cases were stable and low, only about 2, indicating that the first case In the first case, there is a large connection congestion in the connection, but in the second and third cases, there is no connection congestion.

Logon Attempts/sec (logon attempts per second) indicator, in the first case, see the first column of the third row of the chart, this indicator is displayed in black, and Anonymous Users/sec is displayed in red, it can be seen that the change of Logon Attempts/sec is related to Anonymous Users/sec is basically synchronized, but the magnitude is lower than the latter. But in the second and third cases (see the second and third columns in the third row of the chart), the change of Logon Attempts/sec basically coincides with that of Anonymous Users/sec (the line of Logon Attempts/sec covers the line of Anonymous Users/sec line). In the first case, the number of users who try to log in per second is lower than the number of users who connect per second, indicating that there is a waiting phenomenon for login, but in the second and third cases, the number of users who try to log in per second is equal to the number of users who connect per second, indicating that there is no A login wait occurs.

Connection Attempts/sec (connection attempts per second), in the first case, this indicator basically coincides with Logon Attempts/sec, indicating that the connection attempt is more obvious, but in the second and third cases, the indicator is stable at 0 (The lines are black and coincide with the horizontal axis, indicating that no connection attempts occurred.

It can be seen from the analysis that when the website is queuing and congested, its Current Connections is at a high value, and the change trajectory of Logon Attempts/sec and Anonymous Users/sec is similar, but the magnitude is lower than the latter. Connection Attempts/sec and Logon Attempts/sec Coincidence, Anonymouse Users/sec changes drastically and irregularly. When there is no queuing congestion on the website, Current Connections is stable at a low value, Logon Attempts/sec overlaps with Anonymous Users/sec, Connection Attempts/sec is stable at 0, and Anonymous Users The /sec change trajectory is smooth and gentle.

It can also be seen from this that the website has sufficient capacity to handle normal business, and there is no queuing congestion.

An important indicator to judge whether a website is congested is Current Connections.

Guess you like

Origin blog.csdn.net/jhycjhyc/article/details/126639857