Load balancing method in server cluster, as well as node server and cluster

Abstract

The embodiment of the invention discloses a load balancing method in server cluster comprising the following steps: acquiring an address resolution protocol ARP request message sent from a client; if a node server has an ARP response token, returning the client ARP response message, wherein the ARP response token is used for representing the response authority of the node server to the ARP request message; acquiring the load information of each node server in the server cluster, selecting the node server with small load from the server cluster according to the acquired load condition, transmitting the ARP response token to the node server with small load, synchronizing the load information of the node server with the ARP response token to the node server with small load and storing the load information on the load table of the node server with small load. The node server in the embodiment of the invention can flexibly distribute the load according to its own network adapter configuration condition and load condition to improve the server network bandwidth utilization rate, especially to fully utilize all server network adapters and to avoid the performance bottleneck caused by excessive reliance on the load balancing equipment when the server configures multiple network adapters.

Description

technical field [0001] The invention relates to the field of network communication, in particular to a method for realizing load balancing on a server cluster, a node server and a cluster. Background technique [0002] A server cluster is a device that gathers many node servers with the same function to provide a certain service, and provides a cheap and effective method to expand server bandwidth and increase throughput. The effect of server cluster load balancing largely determines the quality of service to clients. [0003] Existing load balancing technology often adopts load balancing device (or proxy server) technology. The server cluster uses a network address externally, and the proxy server maps the network address with a server network address in the cluster according to the load balancing strategy. When the client After the client accesses the network address request to the proxy server, the proxy server maps the network address to a server network address in the ...

AI Technical Summary

Problems solved by technology

[0005] Through the research on the existing technology, it is found that the load balancing device has many single point failures. In addition, since all client service requests will pass through the load balancing device, when there are many requests, the load balancing device will become a performance bottleneck, and in the local server cluster For example, when the NAS cluster file system provides services, the network bandwidth is limited by the network bandwidth carried by LVS; in addition, the way of mapping or forwarding brings additional transmission delay, and in some delay-sensitive applications, it reduces the implementation User experience during load balancing

Especially in applications that require high-bandwidth server clusters, the number of node servers in the server cluster is large, the number of network adapters configured by the node servers is large, and the number of network adapters between each node server is different, making the existing load balancing method Unable to fully utilize all node server's network adapters

Images