In part one of Bass-ackwards storage, I described the tendency of many companies to field storage infrastructure...
before doing their homework on the applications that they are supporting.
Without knowing data access requirements, update frequency and other factors that impact how data needs to be hosted, it is impossible to buy the right kind of storage for your apps. A funny, but true, story was provided to illustrate this point.
In this post, I'd like to suggest five steps that you can follow to right-size your storage decisions -- to align your storage products with the business processes and applications that they must serve. There is no secret methodology here. Like so many things, the steps are derived from common sense.
1. If you are tasked to purchase storage products for your company, it is your responsibility to ensure that you understand business processes and their supporting applications, and the data that these applications use and produce, so that you can define a realistic set of infrastructure requirements. Only once you have these can you effectively shop for a product.
Accept your responsibility and, regardless of your disdain for 30-year-old spinny things, don't outsource your thinking process to any vendor. If you don't do this, you will quickly find yourself in the throes of bass-ackwards storage.
2. Take it as a cardinal rule that the blame for bass-ackwards storage infrastructure eventually rolls downhill to you -- even when you are encouraged by name-brand vendors or analysts to choose a certain approach. So, protect your backside and eschew "marketecture" (marketecture is what happens to architecture when the vendor marketing department gets hold of it).
Case in point: FC fabrics. FC fabrics (oxymoronic SANs) aren't bad choices for all applications, of course. But I will go out on a limb and say that they are probably inappropriate for the lion's share of apps that they are deployed to serve today. Thus, you would be wise to ignore the 'FC SAN-everything' hype. Take it on principle that most applications are more than adequately served by IP storage, or NAS, or direct attached storage.
Also, keep in mind that most FC SANs have not delivered on their fundamental value proposition, except when their deployment conformed to a very narrow set of parameters (for example, when they are created from a single vendor's or vendor cadre's gear). Contrary to popular belief, but well known to anyone who has fielded them, FC SANs do not always deliver non-disruptive scaling, or performance improvements, or better-than-server-attached-storage availability, or enhanced data protection. On the last points, consult the Disaster Recovery Guide 2005 for an interesting survey that identifies FC SANs as the number three cause of application downtime…just behind natural disasters and WAN outages. For the record, server failures have dropped to number four, which calls into question once again the value of "liberating storage from server captivity."
Whether or not you go with a FC fabric SAN, you will be held responsible for the decision. Keep in mind that buying one just for the sake of having one (or because someone said it was strategic) is bass-ackwards thinking.
3. Do your homework on your existing infrastructure. Learn how your data is hosted today and put together a spreadsheet to calculate how much it is costing you to store data on each and every storage device in your shop.
To do this, you will need the current depreciated asset value of each storage platform, plus information on maintenance and warranty costs, and annual software licenses. This info is generally available from your company's asset management system or accounting department. Once you have this data all neatly arranged by storage platform, factor in all of your storage-related downtime this year (there are a lot of ways to calculate what an hour of downtime costs, just search the Web). Then, calculate what you are paying techs to administer and manage your storage hardware and software, and sort these costs by the amount of labor hours associated with each platform.
After a relatively straightforward bit of math, you can derive a pretty good estimate of the cost per GB of each array you have on the floor. That will help you optimize your current storage investment by moving data around to the home that best suits it on a price/performance basis. If you delete stale and contraband data along the way, you may be surprised to find that you have a lot more spare capacity than you thought. That might enable you to forestall new storage acquisitions altogether, reducing the likelihood that you will deploy any bass-ackwards storage.
4. Ask questions. 'Never trust nobody' was the law of the Wild West and today you are dealing with the Wild West of storage. Never accept anything that a vendor says as fact.
Read the technology news pages, the blogs, the stock trading sites and become conversant in your options before you buy anything. Do what you can to find out what others think who have used the product -- both good and bad.
And be sure to share your experience with products on the storage message boards and blog sites that seem to be popping up everywhere these days. This activity is more or less anonymous, so you can feel free to violate the gag orders that EMC has built into its warranty agreement or Network Appliance has just added to its software license that try to stop you from telling the truth about the performance you have received from their gear.
At the same time, use your head. A lot of boards and blogs include items from vendors masquerading as users. When you find what you think is disinformation, bring it to the attention of the site's host or Web master. Or just flame the poster and demand his or her bona fides.
Bass-ackwards storage, when it isn't caused by conscious disregard for proper architecture, is almost invariably the result of poor information to guide decision-making. Remember, Adam Smith's unseen hand only works when consumers make rational decisions based on good information, which is one reason why a storage vendor's stock performance has nothing whatsoever to do with the quality of its products.
5. Make the vendor prove its claims. Times are hard in the storage industry right now: That means that consumers are in the driver's seat. Tell the vendor you have no intention of buying his product unless you can try it out. Don't settle for less: It's your company's money and your job that are on the line.
Testing is the final line of defense against bass-ackwards. Test everything, then test it again -- just to be sure.
That's it for now.
About the author: Jon William Toigo is a Managing Partner for Toigo Productions. Jon has over 20 years of experience in IT and storage.