Server load equalization method for implementing weighted minimum linked allocation

Abstract

This invention discloses a weight minimum joint number dispatch server load balancing method. The technique scheme of the invention are the following: to preset the state representing each server current connection information in the server group, wherein, the said state is represented by data structure with at least one data item comprising data representing the current load states of each server; the said current server load states is represented by the first data item in at least one data item; when load is evenly dispatched, it goes through all first set data items representing current server load states ; it selects server relevant to minimum first preset data set to switch in and at the same time to alter the data item.

Description

technical field [0001] The invention relates to multi-server load balancing technology, in particular to a server load balancing method for realizing weighted minimum connection distribution. Background technique [0002] With the rapid development of Internet technology, Internet users have increased dramatically. At the same time, in order to meet Internet users' higher-level requirements for Internet content and form, Internet content providers use various technologies and means, especially multimedia technology, to make the visual effect of the website more attractive to users. The increase in the number of user visits and the increase in the traffic of a single user visit puts forward higher requirements on the ability of the server to withstand concurrent visits. Since the load capacity of a single server is always limited, the central processing unit (CPU) and input / output (I / O) of a single server will soon become bottlenecks in the face of increasing user access num...

AI Technical Summary

Problems solved by technology

This method only calculates the connection status of each server once in each timer refresh interval, instead of calculating when each connection is added or disconnected, so the calculation results of the server group cannot be refreshed in time

In the case of a large timer refresh interval or a large number of users and heavy traffic, it may cause a sharp increase in the load of a server within a timer refresh interval, or even exceed the load capacity of the server and still be used by the server group. Used as a scheduled server, causing a large number of new access requests to wait for a long time or even be discarded due to timeout

Although this problem can be improved by reducing the refresh interval of the timer, because the fixed-point processor performs division operations slowly and has poor performance, the reduction of the refresh interval of the timer is limited, so it cannot be solved fundamentally question

Images