Method and system for diagnosing Web application performance problem

Abstract

The invention discloses a method and system for diagnosing a Web application performance problem. The method comprises the following steps that filter, capture and deep packet analysis are carried out on the Web application data streams of a specific user through collecting equipment at a certain intermediate point or on terminal equipment of the Internet; TCP and HTTP layer QoS index measurement is carried out on the data streams of one-time Web application session, and TCP connection building Td, TCP retransmission Tr, client side TCP low window packets Tlw, HTTP downloading rate Ht, HTTP first response time delay Hd and HTTP return codes Hc are measured; correlation analysis is carried out on the indexes, and therefore whether the Web application performance problem is caused by a Web server or an IP bearing network or the terminal equipment is judged. The Web application performance problem can be automatically judged and prompted, and common network maintaining personnel can check network faults fast and conveniently.

Description

technical field [0001] The invention belongs to the technical field of network communication, and relates to a method for diagnosing performance, in particular to a method for diagnosing performance problems of Web applications; meanwhile, the invention also relates to a system for diagnosing performance problems of Web applications. Background technique [0002] In recent years, Web applications based on the HTTP protocol have increasingly become the mainstream of the Internet. According to statistics, more than 90% of network resource access and operations (including search, online shopping, online games, online video, social networking, etc.) are all done through web browsers. It can be said that web applications have become an indispensable element in the daily life of the majority of netizens. However, when people use web applications, they often face problems such as unresponsive pages or slow access speeds. [0003] Usually, when netizens encounter the above-mentione...

AI Technical Summary

Problems solved by technology

[0006] (1) There is a blind area in the coverage of the active dial test probe. Due to the huge number of access points of broadband users (often in the order of tens of thousands), it is impossible to deploy the probe to all user access points in many cases;

[0007] (2) There is a lot of arbitrariness in the websites visited by users, and the active dial test system often only monitors the performance of mainstream websites, so the system cannot accurately diagnose the cause of the failure of the performance problems of other websites visited by users;

[0008] (3) The dial-up test system generally monitors different web servers by means of regular polling and dial-up tests. The sampling data obtained by this monitoring method is time-discrete (for example, all the web servers under test can only be completed every 1 hour). Polling monitoring), so there may also be data sampling blind spots for users' fault complaints at a specific moment

Since the protocol process at this stage is relatively simple, there is no specific QoS indicator to measure

Images