Clustering

Anybody working on a network can appreciate the fact when a
server goes down the effects can be catastrophic. Productivity goes for a toss, shared
printers are no longer available and if the server is also a mail server your primary
means of communication to the outside world has just bit the dust. Servers fail to respond
due to a crash, software upgradation or because your hardware has just got cooked. This is
where Server Clustering comes to the rescue.

What is Server Clustering?

Clustering refers to linking multiple servers in a way that allows them to operate as if
they were a single machine. When a server fails another one in the cluster takes over its
duties (this process is known as fail-over). The users will only feel a momentary pause.
The new server in action will cater to the needs of the users. By this time, the server
that has crashed can be repaired and put back to work. Clustering was pioneered by Digital
Equipment in the 1980s.

Here each server can operate on its own ensuring cost effectiveness. The server that
takes over from the machine that is down does its work as an additional responsibility. In
other words the backup server has its own assigned duties but takes over in a crisis.

Types of Server Clustering

Server clusters come in various flavors

- High-availability clustering

Here a second computer is connected to the primary computer to act as a back up. As soon
as the primary system fails, the second computer starts executing everything that the
primary was doing.

- Performance clustering (parallel computing)

Here the computers are linked together to team up and finish tasks quicker instead of
letting a single computer work on its own.

Benefits of Server Clustering

- The failure of the server is easily detected and another server takes over.
Users experience only a momentary pause in service.

- Cluster resources can be surveyed and the workload can be easily moved onto
different servers within the cluster. This helps in manual load balancing.

- While the take off to the new server from the dead one occurs and when rolling
updates on the servers are made, important data and applications can remain online.