Database Clustering

Database clustering is the way of achieving higher availability of database.

DB clustering is a failover technology and not scale-out technology. It does not focus on performance or distributing the traffic to different servers. It merely focuses on moving the execution environment and services to another server in a cluster in case of server failure and thus ensures higher availability of database.

Terminologies:

• Cluster: Cluster is a group of computers configured to work together to serve clients in a similar fashion.
• Nodes: Each server participating in a cluster is called as node.
• Max nodes in a cluster: In Windows Server 2003, the maximum number of nodes can be 4 (with at the most 16 SQL instances) whereas in case of Windows NT, maximum number of nodes can be 2 in a cluster.
• Heartbeats: The nodes in a cluster remain in constant communication through the exchange of periodic messages, called heartbeats.
• Same OS on Servers: A cluster cannot be made of nodes running Windows Server 2003 Enterprise Edition and Windows Server 2003 Datacenter Edition. The OS installed on all the nodes must be same.

Virtual IP (VIP): The client system communicates with the DB server using virtual IP. MSCS takes care of redirecting the client request to the active server and hence the client doesn’t have to worry about which server in a cluster is active.