Are all those new data storage technologies on the market actually new or is it déjà vu, only this time with cool new names masking the old ideas lurking underneath?
A couple of suits in my closet have very wide lapels. I also have a few with skinny lapels. Given the limitations of men's fashion -- especially for "dress-up" clothes -- one of those styles is bound to come back, right? However, I did deep-six the Nehru jacket a long time ago, as I'm pretty sure that one will never come back.
Style is a cyclical thing that goes beyond the world of fashion, so far beyond that it even affects data storage technologies. (You saw that coming, right?) If people lost interest in something four or five years ago, maybe if you just dust it off, make a tweak here or there, and give it a new name, it might just come back into vogue.
Here's a classic old fashion/new fashion storage story. Hierarchical storage management (HSM) was a mainstay for mainframes but a bust for open storage systems implementations. It was largely manual, didn't solve a well-defined problem and was even kind of hard to say. But the idea of putting data on the most appropriate storage based on its value made sense, so a few intrepid storage vendors gave it a nip and a tuck, redubbed it information lifecycle management (ILM) and trotted it out for a second try. The products were a little better than their HSM ancestors, had a mostly new acronym and the word "lifecycle" had a healthy, whole-grain feel. But ILM was a bust, too. I remember trying to put together a story on ILM and not only couldn't we find anyone doing it, it was hard to find anyone who knew what ILM meant.
Fast forward to 2009 or so, when solid-state storage started its incursion into storage systems in corporate data centers. The stuff was so expensive back then that companies looked for ways to make just a little bit of it do miraculous things to speed up storage for critical applications. Once again, enterprising vendors resurrected the old HSM/ILM idea of moving data based on its requirements. In the case of solid-state storage, the requirement was pure unadulterated performance. HSM/ILM principles were a perfect fit: just move data into solid-state when it needed to be there, and get it out and back on cheap hard disks when it no longer needed lickety-split processing. It was a great idea built on "classic" HSM/ILM principles, but what it really needed was a new name: auto-tiering. I'm not sure if auto-tiering is the wide lapel or narrow lapel version of those older technologies, but it seems a lot more stylish.
We've seen a similar swing in styles with cloud storage. It's been around for ages (nearly 20 years), but the services had mundane names like managed storage provider (MSP). MSP fit into that era's xSP naming conventions, alongside Internet service provider (ISP) and application service provider (ASP). There were probably other xSPs I'm forgetting, but whatever they were, they all seemed to blur together.
But that's in the past. Today, we have cloud storage, which is fun, fluffy, ephemeral and always depicted with a cute little picture. No more of that xSP nonsense. You may still not actually know where your data is stored, but hey, it's the cloud and isn't that pretty cool?
Want another back-to-the-future example? How about software-defined storage (SDS)? Besides being a phrase and a concept that stretches credulity (What's been defining storage all along?), it seems to have about a zillion different definitions depending on which SDS vendor you're talking to. The best part of the SDS concept is that it doesn't eliminate hardware; it just kind of adds another layer between you, your apps and your storage. When EMC recently rolled out ViPR -- its version of SDS -- the firm was quick to point out it probably wouldn't mean that you wouldn't need a VMAX or VNX or two (or three) to run under (alongside?) the ViPR software. So it doesn't sound like the ViPR software is doing a lot of storage defining, does it?
That's because SDS is, for the most part, storage virtualization warmed over. Well, maybe not even warmed over all that much, as some of the new stuff lacks the sophistication of storage virtualization products that have been on the market for years. Marketers want to make SDS sound new and exciting, and suggest that hardware will somehow just disappear. But SDS is just like storage virtualization technology that adds a new layer. Granted, it's a layer that should make managing diverse storage systems easier, but it's something extra that ultimately doesn't eliminate anything.
Storage virtualization never caught on big time either. Our Purchasing Intentions surveys show that approximately 29% of companies have virtualized at least some of their storage, with the emphasis on "some," as only 22% have virtualized all their block storage and 23% all their file storage. But those numbers aren't real shockers, even given the obvious benefits of virtualizing storage, especially in mixed-vendor shops. Storage virtualization was pretty expensive, took a lot of time to implement, and was tough to backtrack on and reverse if the results weren't satisfactory. So, with a few exceptions, the storage virtualization market sputtered and stalled for most vendors. That is, until it was reborn as software-defined storage. But let's not rush to judgment: SDS is still a brand-new marketing pitch and we don't know if it'll have the kind of success cloud and auto-tiering seem to be having.
I'm not knocking those storage vendors who pull these old storage technologies out of mothballs, spiff 'em up, give 'em new names, and trot them out as the latest and greatest. It's a good thing, as it shows that good ideas may hang around long enough to get a chance at a second or even third life. And no matter how much the marketing wizards buff them up and re-label them, they'll succeed only if the ideas and technologies lurking underneath the snappy new looks will bring them back into style. Hmmm, maybe I should have held onto that Nehru jacket after all.
About the author:
Rich Castagna is editorial director of TechTarget's Storage Media Group.
- Data Protection Strategies in the Era of Flash Storage –Rubrik
- Data Management Strategies for the CIO –SearchDataCenter.com
- Data integration strategy: A clearer path for data –TechTarget
- Data integration strategy: A clearer path for data –TechTarget