-->

Load balancing

http://www.martijn.com is a nice site to find more info on linux clusters, load balancing (lvs), high availability (ha) and shared storage solutions.

It has a lot information on load balancing, high availability, redundancy and shared storage.

Load balancing is the art of distributing tcpip requests to a pool of real servers. These might be directed to all kinds of services, tcp or udp. Mostly it is used for load balancing http servers, webservers. But mailservers, ftpservers and other internet servers can also be load balanced. Load balancing is made possible by the ip_vs module in the linux kernel. Several Open Source applications use the module for easy to use suites like the Piranha suite from RedHat or the independent Ultramonkey suite. Load balancers are often combined with high availability software because by using only one loadbalancer a Single Point Of Failure (SPOF) is created.

High Availability

High availability can be achieved with heartbeat, or pulse. Pulse is dedicated RedHat software, though portable to other distributions. Heartbeat is available on many distributions. Shared storage like shared scsi drives is also a form of High Availability. Shared storage is a scsi disk or raid that can be accessed by two servers. One form is that one server at the time can access the shared scsi disk. An other form might be the usage of OpenGFS, which allows for symultanious access.