Have we all determined that storage virtualization is good? What do we base it on? I think we all understand the concept: storage virtualization, basically is the process of abstracting logical storage from physical storage where we are presented logical storage for our applications and, through an abstraction layer the virtualization system handles where the storage will reside physically.

Now my biggest question is again: Is this good or bad? Do we have the knowledge of the "backend" system available to us that allows us to make a good decision on an implementation? Do we have the technical "know how"? Do we understand fibre channel, NAS, clustering, data copy techniques, and performance implications?

This is a very serious concern for me. Consider a storage vendor I did some work for a while back. They wanted me to run some benchmarks on their storage for a database system. They told me that data would be spread out across many many disks for load balancing and performance. After two weeks of testing this, I came to the conclusion (with proof) that their system did not actually load balance under all conditions. Basically it didn't load balance for this particular database environment. I actually had to "fake" the system by very unconventional methods to get it to load balance.

Another quick story. Another storage vendor I worked for had an issue with creating snapshots (copies of changed data) for an underlying database when a particular database option was being used.

The moral of the story is plain "buyer beware". There are different virtualization techniques out there and unless you are able to test them under YOUR circumstances, it may not be a fit. Make sure you have the internal knowledge that can adequately stear you away from landminds in storage virtualization.