The role of storage blades in virtual environments

Server blades get their fair share of media attention, but you don't hear much about storage blade technologies and products. Jeff Boles, senior analyst and director, validation services at analyst firm Taneja Group, explains how storage blade technologies fit in today's data centers, and how the technology works with server and storage virtualisation, as well as with Cisco Systems Inc.'s new Unified Computing System (UCS).

Q: What are storage blades and when might they be good options?

A: Today we find customers turning to blade servers to reap the economics of denser, more optimally designed solutions. In general, blades hit the sweet spot between traditional, commercial, off-the-shelf rack servers and the extreme version of optimized rack servers that you might find at Google. Since enterprises can't customize their own hardware and software to the degree that Google can, blades give them better density, performance, and power and cooling than typical commercial off-the-shelf rack servers.

Q: How do the storage blade solutions work with server and storage virtualisation technologies?

A: It varies. Ultimately, when you turn to something like blade storage, you're stepping back to direct-attached storage. And you're giving up a lot of features. That's why we see extremely high attach rates with server virtualisation customers, like 85% attach rates to a SAN [storage-area network] from customers that adopt server virtualisation. That being said, there are some innovative solutions out there that deliver shared storage, like Verari's Open-E stuff, or HP's All-in-One stuff -- although I haven't found anybody that's running server virtualisation with [VMware's] HA [High Availability] or VMotion on top of HP's All-in-One 1200. I'd encourage anybody that's doing so to contact me to say it can be done.

Q: Is there a steep learning curve if an organisation moves from a traditional NAS [network-attached storage] or SAN environment to a storage blade configuration?

A: Sure, it has massive architectural impacts. It's going to change how you do things around those two different types of architectures. Going to blade storage is going to trade-off a bunch of the benefits you get from centralized storage, and you're going to have to look at racking some other management processes around that direct-attached storage. In either case, you need to ask yourself: Is the workload that I'm trying to implement on this blade server appropriate for blade storage? Is the workload you're trying to put on a blade the right workload for blade storage?

Q: How does Cisco's UCS architecture change the landscape for storage blades?

A: That's a great question. I think Cisco's Unified Computing System technology is a bit of a paradigm shift for blade servers because they've unified the I/O coming out of the blade server, so now it's a one-wire solution. And that tackles one of the sticky problems for traditional blade servers that necessitated blade storage. It was too hard to get all that wiring and connectivity to the blade servers in many cases, and it drove up the expense. By taking the I/O and SAN connections outside of the blade server itself, Cisco has made it much more cost effective to use network storage, whether it's SAN attached, NAS or iSCSI. It becomes really practical and compelling to look at SAN storage for all your server blades.