On July 25, 2007, I published a storage article entitled, "Shared block-level storage continues to become more accessible." In response to my statement: "A few years ago, the idea of a Fibre Channel-based SAN didn't even get raised because of the cost and complexity of such a solution. It was RAID all the way in most servers," TechRepublic reader gintas@... commented:
"Pardon me, but how come SAN and RAID are mentioned there in same level? What do SAN and RAID have in common to achieve same goal? Is SAN of any type interchangeable with RAID of any type?"
I wanted to take a little time to discuss my reasoning for this statement, provide a clarification, and share my thoughts on how I believe many IT managers in small and mid-sized shops view storage. By the way, when I talk about RAID in this article, I mean the direct-attached kind. I understand that SANs also use RAID, but that isn't the level I'm talking about here.
I don't think that SAN and RAID are necessarily interchangeable when it comes to storage. Rather, I see SAN as the next step in the "storage maturity level" for many organizations. Most organizations follow a path that is similar to this:
1. Need space. Individual servers with direct-attached storage (no RAID).
2. Need for additional protection and availability. Move to individual servers with RAID storage.
3. Need better storage management and efficiency, virtualization/DR, etc. Move to shared block-level storage and away from direct-attached storage.Figure A
An extremely simplistic look at the phases of enterprise storage solutions
It's probably important to note that this simplistic look at storage very likely closely maps to an organization's reliance on technology for day-to-day operations, so "storage maturity" may be a misnomer ("technology reliance" or "technology integration level" might be good, too), but fits as a way to frame the context of this discussion.
When I indicated that, for many organizations, it was "RAID all the way," I meant that the cost and complexity of a shared block-level storage solution, or some kind of SAN, was beyond either the technical or financial means for many organizations, thus leaving these organizations at the second level of storage maturity. In fact, a SAN is still out of reach for many organizations, but this is mostly due to a lack of funds, since today's Ethernet- and IP-based storage solutions have taken much of the technical complexity out of the solutions.
Now, there are a ton of cheap SAN solutions out there today, but that doesn't mean that even a cash-strapped company would jump on an "el cheapo" SAN just to get one. In many cases, I really think that organizations with a need for a SAN want to do it right. They want to make sure that, beyond raw storage, they are getting the features that they need, such as replication and snapshots.
Ultimately, although a SAN may have some whiz-bang features and enable new capabilities, such as enhanced disaster recovery or better virtualization, at the end of the day, SAN and RAID (and even direct-attached storage), are there for the same reason: to store organizational data. In that context and in the context of describing the phases an organization's storage maturity, RAID and SAN are on a level playing field, but each much be matched appropriately to the needs of the organization.