Method and system for ensuring application service quality in distributed environment

Abstract

The invention provides a method and system for locating bottleneck nodes and guaranteeing application service quality in a distributed environment. The method for locating the bottleneck node includes calculating the delay fluctuation value of each node on the critical path of the service in its processing stage; and determining the bottleneck node according to the delay fluctuation value. Among them, the critical path of the service is obtained according to the critical path of processing the request for the service within a certain period of time; the delay fluctuation value is obtained according to the time when the node processes the request in its processing stage within a certain period of time. The method of guaranteeing application service quality includes locating bottleneck nodes for services with long-tail delay; and checking whether the delay fluctuation value of the bottleneck node exceeds a predetermined threshold, performing fault diagnosis according to the inspection result or performing request scheduling or request scheduling for service requests of the bottleneck node accelerate. The invention reduces the fluctuation of the request response time and the long tail delay, and also reduces the overhead of optimizing nodes level by level.

Description

technical field [0001] The present invention relates to the technical field of distributed computing, and more specifically, to a method and system for locating bottleneck nodes and guaranteeing application service quality in a distributed environment. Background technique [0002] Currently, Internet applications such as e-mail, search, online shopping, social networking, online video, and network maps have become a part of people's lives. These applications often serve hundreds of millions of users, which means that Internet applications have become social public services like electricity, and various distributed computing environments that support Internet applications with a large number of users have also become social core infrastructure like power plants . [0003] In a distributed environment, generally, Internet applications are divided into many service nodes (or service modules) for step-by-step processing, and these service modules are deployed on multiple serve...

AI Technical Summary

Problems solved by technology

However, this method will dynamically adjust the priority of each node on the request delivery path, and is only applicable to the dependent / serial mode. When applied to the division / aggregation mode, there will be relatively large overhead

Images