filograph - Fotolia
- Rich Castagna, Vice President of Editorial
In a blizzard, it's hard to see a single snowflake. And with the avalanche of new data storage technology that has swirled around data centers the past couple of years, it can be pretty tough to pick out that exemplary piece of engineering innovation and dexterity. They say no two snowflakes are alike (how "they" came to that conclusion, I'll never know) and, similarly, this data storage maelstrom is marked by a staggering number of new products and product categories. In other words, it's tough to figure out.
If you haven't cast your company's lot with the cloud, software-defined storage, hyper-converged systems or any of the other storage buzzwords du jour, you're not alone. Just as a whipping snowstorm can reduce one's visibility, the flurry of new -- or re-invented -- storage techs that we've been weathering can also make it hard to discern distinct shapes in the storage market.
There are a couple of reasons for these conditions and the uncertainty they bring. As I've noted a few times in this column, there really is a lot of stuff going on in the world of storage these days. There probably isn't one single reason why there's so much activity on so many fronts, but the success and penetration of server virtualization has certainly upset the status quo in data centers. Add to that the ever-growing data stores that most companies continue to stash away, toss in some mobility and cloud economics, and there are plenty of reasons why smart storage people are rethinking the way things have been done and how they can be done better. Also figuring into the picture is the inevitable IT cycle of consolidating operations and services to manage them centrally … until somebody realizes that distributing those processes and services makes more sense. I'm not sure if we're entering the consolidating or distributing phase as there's evidence of both in the new storage lineup, but whatever it is, it's a factor for sure.
Another reason why the situation remains so unsettled and potentially perplexing is that we haven't yet seen a single one -- or even a small bunch -- of these technologies rise to the top to take a market lead or even set a direction. That's partly because some of these new techs look more like niche or specialized products than general-use storage, and because the technologies (in many cases) still have some significant limitations that have held them back from more widespread acceptance.
Let's break things down tech by tech -- assuming hyper-converged systems, all-flash arrays, cloud storage and software-defined storage are the hottest recent data storage technology developments.
Hyper-converged systems. These are more than just storage, of course, but some companies are buying them, or at least evaluating them, when they're ready to retire an older array. Hyper-convergence offers attractive advantages such as convenience, ease of use, fast deployment and so on, but many of these systems lock you into a single hypervisor, some have very limited scaling options and, if you plan to hook in physical or virtual servers outside of the hyper-converged ecosystem, you can forget about it; this is not your father's networked storage. (See this month's cover story, Hyper-converged systems deliver storage in a bundle.)
All-flash arrays (AFAs). A lot of AFA vendors brag about how -- through the magic of compression, deduplication and creative firmware engineering -- they've brought down the price of all-flash arrays to just $3, $4 or $5 per gigabyte of capacity. That's still a lot more than hard disk drive-based systems, and sometimes the compression/dedupe ratios used to derive the "effective capacity" are, shall we say, a tad optimistic. As much as AFAs have begun to look like good old enterprise arrays by adding snaps, replication and other big-boy features, their key attribute is blinding speed to feed your most performance-hungry applications.
Cloud storage. Cloud storage is cheap and easy, expands and contracts as your needs change, and it helps you to avoid all kinds of costs like labor and developing specialized expertise. True? Well, sort of. But there are some serious tradeoffs to those benefits. If your capacity actually needs to contract at times, the cloud is a pretty attractive alternative, but in most companies, capacity grows and grows. It's also true that storing data in the cloud helps you to avoid the cost of an on-premises storage system and the power and space it requires. But you'll be paying to keep your data in the cloud forever. So what might seem cheap at first isn't such a bargain when you multiply the costs by months or years or decades. And getting data in and out of a cloud storage service is still the chief bottleneck. Sure, you can spin up virtual machines in the cloud and access that data without actually retrieving it, but that puts your servers in the cloud, too. A lot of companies are still very uneasy about shipping so much of their intellectual property into the ether.
Software-defined storage. A lot of confusion persists around this storage tech market, despite the efforts of vendors, analysts, journalists and just about everybody else to define it. It has a very enticing "we only have to buy software" air about it, but as everyone knows, storage is also about hardware. As the category has evolved into things like VMware's EVO:RAIL software, it becomes clearer that while the software might be doing a lot of the defining, the hardware is still doing the heavy lifting. This product category still needs to sort itself out and be able to present more concrete benefits before it can shake off the niche label.
So, storage still needs a fair amount of sorting out. I have no doubt that each of the above technologies will mature and assume key roles in many companies' storage infrastructures. But until they do, expect those blizzard conditions to persist.
BIO: Rich Castagna is TechTarget's VP of Editorial/Storage Media Group.