Purchasing a storage array to replace, or at least augment, your direct-attached storage (DAS) is an important and potentially expensive investment. Many businesses know they need some kind of data storage rationale before the number of disks get completely out of control and become unmanageable, unaffordable and unsustainable. Knowing what questions to ask before meeting potential vendors will prevent a reseller from pushing you down a purchasing route that might not be in the best long-term interests of your business. Be sure to ask a value-added reseller (VAR) the following five questions before you purchase an iSCSI storage array.
So, I put a shelf full of disks onto my network and my problems are solved? How does that work?
An iSCSI storage area network (SAN) works by placing what is, in effect, another server onto your network. It has memory (lots of it), processors and network cards. The only difference is that the storage controller is optimized for serving up data from disk and storing it to disk. Don't think of a SAN as a magical black box that you can't go near or do anything with.
A reseller should ask you for a lot of information before coming to talk to you. They'll ask you how many file servers and database servers you have, and the amount of disk space you own. More importantly, they'll ask you how many physical disks you have, and the capacity of the disks. Those numbers will allow the potential reseller to offer you the right storage controller and the right number of disks, to provide at least the same level of I/O performance as all of your current servers. If they don't ask you these questions, you're not dealing with the right people.
I have this box that contains all of my data. How on earth do I manage it? Do I have to call you every time I need something done?
Training is essential. The SAN should have an easy-to-understand Web-based interface. While the people who set up your SAN and do the initial work will go straight into the command line, that's only because they can work faster that way. Make sure they show you the interface and how to configure storage so that a server can gain access to it. Forcing you to go into the command line for day-to-day tasks is not acceptable.
The reseller should explain the difference between blocks and files and how the storage controller takes a "photograph" of its own disk map so that it can revert to a previous version of a file or database on that disk. Let them draw on a whiteboard for you; it will help you both.
What are your RAID levels? How do they measure up against the competition, and why do I care?
RAID is all about resiliency. However, if the array spends too much of its processing power building and maintaining RAID, it will affect the potential performance of your SAN. Implementing a SAN brings the 50 or so disks that were directly attached to servers under a single controller. Those disks need to be able to deliver the same performance, i.e., provide the same amount of megabytes in and out as the direct-attached model. Each vendor handles disk resiliency in their own manner. NetApp only allows RAID4 or RAID-DP (A RAID6 implementation). Others allow multiple RAID levels. Don't pick a RAID solution just because you're familiar with it as it applies to direct attached storage. Allow the vendor to explain how their particular solution works and distributes information across the disks.
How far can I expand this storage array before I need a new one, and what are my options when that time arrives?
Once you implement a SAN, you'll want to add more disks to it over time. You'll want to put more servers on it, perhaps run more applications (an Exchange archive solution perhaps) and you'll probably want to keep more information on it. Physically expanding the disk array must be as simple as adding a new shelf of disks, connecting them and switching them on -- all without turning the SAN off at any point. Expanding storage must be as simple as making a few mouse clicks to bring the new disks into the existing space. Don't accept any expansion solution that means you have to switch anything off. All the configuration information should be stored on the existing disks.
About the author: Mark Arnold, MCSE+M, Microsoft MVP, is the principal consultant with LMA Consulting LLC, a Philadelphia, PA-based private messaging and storage consultancy. Mark assists customers in designs of SAN-based Exchange implementations. You can contact him at firstname.lastname@example.org.