Managing storage technology is an oxymoron these days. It seems like the more tools you add to manage your increasingly complex environment, the more convoluted the environment becomes. Nirvana is within sight, but your budget prevents you from getting there. Even worse, the budget shrinks but the demand on your resources grows. What's a storage area network (SAN) guru to do?
As you choose different technologies, a Pandora's box of complexity is opened. The more solutions you invest in, the easier life is supposed to become, but now you are managing more and more moving parts, increasing business risk. The converse of this paradox is that the less you invest in tools, the more single threaded (KISS principle) and expensive your solution. Using a one-size-fits-all approach isn't very cost effective. Can you find a balance somewhere in the middle?
The answer is finding that line between new tools and keeping things simple. Here are a few tricks to make the balancing act easier.
First, define your goals. You must have a clear understanding of what upper management is looking for in terms of results. Once you understand their goals, the rest is as easy as brain surgery (in the off chance you are a SAN manager and a brain surgeon, it's as easy as rocket science).
The second trick? Knowing you are always going to have to do better tomorrow than you did today. Acceptable performance four months ago will not be acceptable four months from now.
It is a given that your IT environment will continue to grow. Even as companies strive to do more with less, they continually invest in and rely on IT. It's the unwritten rule of our world. CPUs are becoming increasingly more powerful, but the software that runs on them is also becoming more CPU intensive. RAM is now inexpensive and plentiful, but the applications we are running chew through it like candy. SAN drives are getting bigger, arrays are more powerful, cache is more intelligent, but the demand for storage will never slow down. As a result, look to buy big.
But, there's a difference between growing your environment and enlarging it. It will always be more attractive to buy the "budget storage array" vs. a larger array with fewer spindles. It is always less expensive to buy a 48-port switch than it is to buy 48 ports on a director class switch. What is not apparent is that in the long run you will spend more time architecting, managing, upgrading, troubleshooting and complaining about a gaggle of smaller switches than you would if you had one large switch. Remember, your environment will grow, and it's much easier to make investments in equipment that can grow with it than to bolt on another "budget" stop-gap solution.
The third trick is to arm yourself with as much information as possible. The dozens of SAN infrastructures I've either managed or reviewed all had one thing in common: they were over-designed on purpose and usually over-designed/over-provisioned because of a lack of knowledge about what is truly going on. The lack of knowledge ranges from not knowing what the applications might do throughput-wise to not knowing the throughput requirements. It is a sad fact that we SAN guys don't have all the magical tools that our IP network brothers get -- traffic flows, traffic patterns, trended errors and so on. Oh, to be a network guy!
Knowing when your ever-growing demand from Server 1 actually requires that additional host bust adapter (HBA) is invaluable. Being able to see its traffic pattern (on the SAN), including heavy traffic moments and less intensive periods, can mean knowing you can complete a level 0 backup within the backup window without investing in another HBA (more cables and another switch port).
There are a few products on the market that can give you this type of information. For my money, the best solution I could find was Virtual Instruments' NetWisdom. It is a bit strange at first to have access to the type and amount of information NetWisdom provides. Once you get over the "drinking from a fire hose" sensation, you instantly start to see its power -- trending out usage, actually seeing a non-biased response time on exchanges, or even just showing block sizes on the SAN in real-time is invaluable as you attempt to squeak out a little more performance every day.
Managing your storage area network is a daunting task. It is the fastest growing, usually the most expensive, the loudest, the biggest generator of heat and the heaviest area of the data center. You opened a Pandora's Box of intricacy when you invested in shared storage. Following a handful of simple best practices should make your life a little bit easier. At the very least, your late night pizza and soda consumption will decrease.