What would be the main advantage of choosing a platform, maintained using a SAN, over a platform that is merely clustered?
Any clustered "platform" would need external storage to enable resource sharing between the cluster nodes. Although MSCS is known as a "shared nothing" clustering solution, it actually shares disks between the clustered nodes. This can be done with a standard external SCSI based solution using "Y" adapter cables and external SCSI terminators, or through SAN based disk. The SAN solution is much cleaner and faster though.
There are software only clusters like the original Netware solutions, where the internal disks are used but these tend to be "hot standby" clusters. The best bang for the buck in clusters is to use active/active nodes, which is made possible by "resource sharing" of the external SAN based disks. All cluster nodes can have access to all cluster resource disks, which enables fail over of any resource to any node. SAN-based storage also allows for much greater distance between the storage devices and the host servers. In fact, you can create a "stretched cluster" solution, where the end host nodes are geographically dispersed in different locations and resource failover is automatic between DR sites when a host fails.
Only SAN has the capability to do this today.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our .bphAaR2qhqA^0@/searchstorage>discussion forums.
This was first published in April 2002