The currently available software for monitoring and collecting metrics for your environment typically just wants to know storage allocated versus used, to know how much networked you need to assign. But, important pieces of data are missing: the disk I/O profile of the server, disk queue length, read versus writes, paging, swapping and a variety of other data that can help you make storage architecture decisions. All applications have a performance profile that designates them as either CPU intensive, memory intensive, disk intensive or network intensive, or some combination of all of the above.
If you have a storage intensive application (VM), you will want to consider not grouping those VMs together, just as you would not want to group memory intensive or CPU intensive applications. Another thought to consider: if you group applications or VMs together that have the same recovery requirements, they can be grouped together for more easily replicated storage. RAID configuration impacts performance. Large LUNs are more difficult to backup, replicate and restore. Small LUNs have more complexity, less flexibility and less portability.
This was first published in December 2007