Clustering protects from localized hardware failure (for example a power supply failure on one node), but does not protect from a power failure in the building supplying all nodes in the cluster. Geo-clustering can keep data in sync between two geographically dispersed servers, and should you experience a regional failure (a flood) your applications can reconnect to the DR site.
The challenge for VARs is to:
- Provide servers with hardware redundancy so that they will be as resilient to hardware failures localized to that node
- Provide clusters which will provide high availability so the cluster will be as resilient as possible to node failures (for example a unexpected SQL Server shutdown due to a program fault, or a planned outage for hardware servicing or a service pack). In a cluster you can take a node offline by failing over to other nodes in the cluster.
- Provide DR solutions which will protect the client's data against regional failure by keeping a DR site in sync with a near real time copy of your data.
This was first published in April 2007