Simplicity is a term that I am starting to encounter more often in presentations I attend and in materials I read about storage. This is understandable. Organisations cannot allow storage networks to become excessively complex lest they become unstable. So here is some advice on how to keep SANs simple and manageable.
First, be extremely cautious about connecting SAN islands. Visios that illustrate the value of virtual SANs and interSAN connectivity are one thing, managing and controlling that connectivity is quite another. This is especially true in dynamic environments where HBAs may be moved from a server on one SAN to a server on another that results in a server accessing storage that it should not be connected to in the first place.
Second, introduce change control into your production SAN. Change control requires that SAN changes be announced ahead of time, documented and then double checked. Yes, work will slow down, maybe to the extreme agitation of some overanxious users. But it only takes one corruption of a DB2 database for these same users to become equally upset wondering how you could have been so careless as to have allowed this corruption to occur in the first place.
Finally, limit the number of storage interfaces that servers may access. This can be done in one of two ways: one, virtualise the storage at the network level or consolidate as much of the storage as possible into one large array. These steps limit the number of interoperability questions and, even if only 80 percent successful, help to make SANs easier to manage.
Simplicity is a choice, not an accident. So while implementing innovative features may be technically exciting and save money in the short term, they also introduce levels of complexity and risk that not every organisation is equally ready to manage.
Jerome Wendt currently works as a storage engineer and storage analyst. He contributes regularly to a variety of industry trade publications and can be reached at [email protected]