Not too long ago, I wrote a post "Are Servers The New Storage?" where I made my case that newer software-based storage platforms -- based on commodity servers and agnostic software stacks -- would become increasingly popular for a number of compelling reasons.
It seems that things are moving along very rapidly in that direction. More and more viable options are appearing in the market, with even more coming.
For those of you who are involved in your organization's storage strategy, I'm sorry to say you'll have even more options to pore through going forward.
But it'll be worth it :)
Besides the obvious "inexpensive hardware" argument, I presented four reasons why I believed servers-as-storage would become more attractive.
From a pure performance perspective, there doesn't appear to be anything faster than applications running on server-based memory technologies. That architectural bang-for-buck differential is substantial and will be persistent.
There's also a convergence argument: storage-as-software can be potentially far easier to manage as part of converged operations and workflows. If you've taken a look at VMware's recent VSAN beta, you'll know what I'm talking about.
Extending the economic argument even further, there's a pooling benefit: hardware resources used for any storage purpose can be shared and reallocated: not just among storage tasks, but general purpose server tasks as well.
And then there's a simplicity benefit: using the exact same building blocks (industry standard servers) for both compute and storage.
The New Options
While there have long been various software products that transform raw server hardware into usable storage subsystems, the tempo has notable picked up in the last few months.
One example: the interest in the VMware VSAN beta quite strong, and generated a ton of enthusiasm -- much more than I would have guessed at the outset. EMC's acquisition of ScaleIO fits into this vein, but for entirely different use cases.
Today, Coho Data came out of stealth with an interesting approach, although not entirely hardware agnostic. And there are wonderful established solutions from very focused vendors like Scality that you don't hear much about, but probably should.
Feel free to add more in the comments section, if you think I've missed something important :)
The bottom line? Storage-as-software is shaping up to be a significant category within the broader storage marketplace. And this time, it looks for real.
Change Ain't Easy, Is It?
But broad adoption of storage-as-software won't be without its headwinds.
For example, there is a constant dynamic in IT: standardization vs. optimization. Standardizing on as few technology choices as possible pays real benefits in simplicity, efficiency and re-usability -- strong incentives indeed.
But, on the other hand, optimization (using different technologies for specific use cases) can pay benefits as well: economics, differentiation, etc. The internal debates go back and forth -- the world will never be one way or the other.
When it comes to enterprise storage, the optimization camp is winning, based on what I see. The workloads are too varied, the requirements all over the map, the technology choices very disparate, the aggregate investments growing, etc.
Just take a look at the broad portfolios from any of the major storage players, and you'll see specialization, not standardization. If anything, storage platform diversity appears to be increasing, not decreasing.
The focus then has to move to standardizing operations vs. standardizing technologies -- and there's still more much work to be done there.
How Will The Storage Team React?
Enterprise storage teams tend to be a justifiably conservative bunch -- after all, there's no bad IT day like a bad storage day. You'd update your resume, but you can't get to it.
Now we have to consider a passel of newer software-on-server storage platforms with two important attributes: (a) they're relatively new, and (b) they tend to be owned and operated by the server team as a part of general infrastructure vs. a dedicated storage team.
The storage pros can definitely help with the first part. At some level, storage is storage, and the in-house storage team can make a big contribution in sorting through the newer options, figuring out what makes sense, and generally adding their unique expertise to the discussion.
What won't be productive will be anything that is viewed as a defensive move by the storage team to ostensibly protect their turf. And, unfortunately, legitimate concerns can often have that appearance, so be judicious and tactful :)
And, just to state the obvious, the world isn't going to run on software-based storage array stacks overnight. There are plenty of workloads that will need to run on traditional storage arrays for the foreseeable future, and that's not going to change quickly.
Figuring out how and when to use the right storage technology in the right situation will continue to be a valuable skill -- even if some of those technologies come in the form of software, not hardware.
2014 Is Shaping Up To Be A Fascinating Year For Storage
As I think about 2014, I think it's going to be far more interesting than usual from a storage perspective. The move to flash is now a given -- you can't call it disruptive anymore, it's now just part of the landscape.
Personally, I think we'll see more than a few of the all-flash-array startups fall by the wayside: the herd will likely be culled.
We'll have good storage control planes to work with -- EMC's ViPR comes to mind -- which can deliver the key abstractions we'll need to orchestrate storage in all its forms. Heck, even tape might get interesting again :)
And, perhaps most interesting, we'll have several all-software storage solutions that are very different from the dedicated arrays that preceded them.
Storage never, ever gets boring :)
Like this post? Why not subscribe via email?