摘要 |
The invention relates to the backup of a user session on a network. When the user at a client, such as an ISP, makes a request on the network, this request is routed by a load balancer to a primary host server on the network. The selection of the server is done by the load balancer in such a manner as to distribute the load on the system evenly and prevent overloading of one particular server. Once the user request has been routed, user specific information such as a session identification code or a password is provided to the primary host server and the session is set up. A single session must comprise of at least one application process on the selected primary host server. On commencement of the session, the load balancer designates at least one other server on the network to act as primary backup server for the user session set up on the primary host server. The primary backup server stores session information from the primary host server and in case of a server breakdown, the backup server takes over as the primary server. The load balancer then configures another server to act as backup server for this new primary server. According to the invention, the backup function of the servers is equitably distributed across servers so as to ensure that no one server is excessively burdened with the backup functions. |