This article can also be found in the Premium Editorial Download "Storage magazine: Exploring new options for disk-based backup."
Download it now to read this article plus other related content.
A pretty senior guy at one of the large storage vendors recently declared "Let the virtualization wars begin!" during a phone conversation we were having. If we'd been on a videoconference, he would have seen my eyes roll. It's not like all of you will rush to take sides as EMC, Hitachi Data Systems and IBM roll out their respective virtualization/account control strategies.
The storage industry has a penchant for "wars," most of which remind me of the '60s bumper sticker "What if they gave a war and no one came?" No one but analysts and press came to the SAN vs. NAS war or the Fibre Channel vs. iSCSI war, and I predict that no users will come to the virtualization in the array vs. virtualization in the network war either.
Honestly, I don't have an opinion about which approach is better. I'm pretty sure they all need work. There are persuasive arguments on both sides, but most of those need to be proven in practice. However, I do believe that overheating the debate with apocalyptic rhetoric is counterproductive, if predictable.
Don't buy into it. A more appropriate approach is to start with your basic needs and then work through various scenarios to see how well they are met. As I see it, virtualization has a few virtues that are central to most of you right now, such as:
- The ability to expand volumes easily, even across different arrays because particular applications require that capability.
- The option to control multiple vendors' arrays from a common point and with common management tools.
- The power to replicate and migrate data across various classes of storage, regardless of what array they sit in.
- And last, but not least, more flexibility in continuing to gain some use out of legacy arrays.
Most virtualization schemes promise those things. But the fine print contains some warnings about performance, recoverability and interoperability, just to name a few things. Hard questions need to be asked. Sure, each vendor claims that its approach preserves your investment in other firms' gear, allowing you to fold that into the virtual pool. What about the reverse? Will all of your storage vendors promise to play nice and allow their gear to work in other vendors' virtualization schemes? Are they serious or just offering verbal appeasement?
What happens to your arrays if you want to back out of a particular virtualization scheme? How easy is it to recover from a disaster? Architectural arguments about performance are important, but where's the benchmark data? And will they benchmark their gear in your environment?
Virtualization is a vital piece of reining in the complexity of storage management, and probably the only route toward ending the fragmentation of storage and laying the groundwork for true enterprise storage networks. No one gives a second thought to the notion of an enterprise data network, but we're still a few years away at best from that possibility in storage. Engaging in paintball wars over whose virtualization is best won't get us any closer.
On another note, you'll notice this issue sports a design refresh. We wanted to tweak our look to make articles more accessible and quicker to read. We're looking at this as a point release, a chance to fine-tune things after three-plus years of publishing this magazine. You'll also begin to see that we're adopting more of a common look between Storage and our sister efforts, SearchStorage.com and our Storage Decisions events. This will make it easier and more sensible for you to move among the different types of information that we feature in each venue.
Love it or hate it, please let us know what you think.
This was first published in June 2005