Load balancing method, load balancer and system thereof

Abstract

The invention discloses a load balancing method, a load balancer and a system thereof. The method comprises: when a request form a client is received, a session Id is obtained from the request; and a server corresponding to the session Id is searched in a session list and the request is sent to the server, wherein a server address corresponding to the session Id is preserved in the session list. According to the technical scheme, the session Id is obtained in the request from the client and the corresponding server is determined by using the session Id, thereby ensuring that no interruption occurs during the session process. Moreover, load balancing is realized; and thus problems that session may be lost or the load is not balanced in the prior art can be solved and the beneficial effect of load balancing realization on the premise of session keeping is realized.

Description

technical field [0001] The invention relates to the technical field of computers, in particular to a load balancing method, a load balancer and a system. Background technique [0002] After the introduction of cluster technology, load balancing, as the core of cluster technology, has received extensive attention and research. Load balancing must not only ensure that the load between back-end servers is equal, but also correctly allocate requests for the same session in scenarios that require session retention. to the corresponding backend server to ensure that the session between the client and the backend server is not lost. [0003] LVS (Linux Virtual Server, Linux Virtual Server) is an open source load balancing technology. By inserting the call method at the detection point (HOOK) point of Netfilter (Network Screening System), the IP (Internet Protocol, the protocol for interconnection between networks) ) layer realizes the transparent forwarding of data packets. LVS m...

AI Technical Summary

Problems solved by technology

[0005] The disadvantage of the above solution is that there may be session loss and serious load imbalance in the high-speed proxy Internet environment. figure 1 It shows the session process in the Internet environment of multiple proxy servers. The requests of a large number of client users are sent to the load balancer connection through the ISP (Internet Service Provider, Internet Service Provider) of multiple proxy servers or the intranet of the enterprise, and come from the same proxy server. The two requests of the same session of the client may go to different proxy servers to reach LVS (load balancer). LVS judges that the source IP is different, and allocates the two requests to different back-end servers, resulting in session loss; for example, LVS will The access request from Proxy1 is selected according to the IP address of proxy server 1 using the load balancing algorithm figure 1 The top server is used for processing; and the access request from Proxy2 is selected according to the IP address of proxy server 2 using the load balancing algorithm. figure 1 Bottom server for processing; resulting in lost session

figure 2 It shows the session flow in the single-proxy server Internet environment. A large number of client requests arrive at LVS through a single proxy server. LVS judges that the source IP is the same, and allocates these requests to the same back-end server, resulting in overloading of a single server. Other servers but very free

For example, access requests from three clients all arrive at LVS, and LVS judges that they are all access requests from Proxy1 (the source IP is the same), and the load balancer forwards all connections to a server according to session persistence, thus causing the server to fail. unbalanced load

Images