A distributed system service load balancing method and device

Abstract

The invention discloses a service load balancing method and device of a distributed system. Compared with the prior art, which only performs load balancing according to network port traffic, the invention comprehensively considers the load situation of the service and the load situation of the server where the service is located. After the traffic of each network port corresponding to the service is determined, the busyness of the service is determined, and whether it is suitable to continue using the service is judged, and the master-slave service is switched according to the judgment result, so that it can effectively avoid the problem caused by traffic size alone. The one-sidedness caused by determining the availability of services or servers can more accurately determine the busyness and applicability of services, and effectively control the probability of system overload. Moreover, the present invention only switches over all overloaded main services and standby services on the server, rather than switching over the entire server, thereby ensuring more effective use of all hardware resources and better monitoring of the server. Each service can be flexibly deployed.

Description

technical field [0001] The invention relates to a load balancing technology, in particular to a service load balancing technology in a distributed environment. Background technique [0002] With the further extension and development of industrial control systems, it is necessary to provide more powerful processing capabilities in order to be able to handle the increasing number of user visits and processing requests in a more timely manner. Generally, a series of servers providing the same service content are set up to form a server cluster to respond to the processing request of the client. After receiving a processing request from the client, it is necessary to perform load balancing calculations to determine which server to send the request to for processing. [0003] The distributed system provides users with a rich collection of resources, so it is increasingly widely used and is valued by people. In the existing technology, the load balancing calculation is generally ...

AI Technical Summary

Problems solved by technology

Furthermore, the inventors of the present invention found that the size of the traffic has a very one-sidedness to the load of the server or computer, and some services have less information interaction, but the calculation is complicated

In view of the individual performance differences between servers and the differences in service resources consumed, only relying on network port traffic for load balancing calculations cannot achieve a true adaptive load balancing service scheduling mechanism, and cannot efficiently utilize all hardware resources. Effectively control the occurrence probability of system overload

[0004] Moreover, the existing technologies generally perform load balancing based on servers, that is, during the load balancing process, the active / standby switchover is performed on the entire server. The overall switching based on the server will switch all the normal or overloaded services on the server to another server, which will easily cause the other server to be busy

And only one of the active and standby servers is enabled at the same time, and the existing hardware resources are not fully utilized

Images