Greg Blomberg - Fotolia
I once thought that data storage was a straightforward technology, with fairly logical alternatives. To quote Bob the Bard, however, "I was so much older then. I'm younger than that now." When I look around at the current data storage landscape, I wonder how I could've been so naïve. This stuff is getting pretty damned complicated.
I've written about how the number of options in the data storage world has grown precipitously in just the last few years, with more alternatives than ever for whatever storage problem you're trying to solve. The variety of data storage technologies runs from media to array designs to software-centric approaches to entirely new architectures. I first wrote about this new data storage landscape a while ago, and it has only grown more complex -- further underscoring storage's evolution as a genuine good-news, bad-news storage story.
It's nice to have choices. That's the good news. But too many choices can be bewildering. That's the bad news -- at least, part of the bad news. Many of the new data storage technologies, once they're rolled out onto the data center floor, may have repercussive effects on other parts of the storage infrastructure. It's a kind of whack-a-mole effect that can cause new bottlenecks to pop up after eliminating an old one.
Changes in the storage market
Stepping back a bit, it looks like the data storage industry is still trying to adjust to its newfound "solutions." IDC recently reported that worldwide sales of enterprise storage systems dipped 7% year-over-year (to a trifling $8.2 billion). Measured by total capacity shipped, the industry suffered a 4% downturn. Among all the storage heavyweights, only Hewlett Packard Enterprise registered any growth in revenue for the first quarter of 2016 (11%).
At a time when big data and the internet of things are beginning to dominate many IT conversations, it seems inconceivable that companies are actually buying less storage. But they are. And you can go back to all the new data storage technologies to see why that's happening. Companies are probably storing more stuff than ever -- that curve in the capacity graph never bends downward -- but they're doing a much better job of it and doing it more efficiently.
Impact of flash and hyper-converged storage
For example, all-flash arrays (AFAs) are typically considered purely a performance play -- and why not, as they run circles around spinning disks. But when a company peps up the performance of a critical app by moving it from a disk-based array to an all-flash system, there will likely also be an installed capacity reduction. Short-stroked to squeeze out the last bit of performance, the capacity of the old hard disk array was probably way overprovisioned. The new AFA delivers its sizzling performance without having to overprovision and, in fact, because it can even spare a few cycles to dedupe its data, the flashier array might need much less capacity.
Similarly, we can see how new architectures like hyper-converged infrastructure can introduce more efficient use of storage because of its modularity and easy scaling. The rebirth of server-based storage built on PCIe and nonvolatile memory express flash has also helped reduce capacity requirements by focusing capacity and performance specifically where it is needed.
The Evaluator Group, in conjunction with the Storage Networking Industry Association, recently published the results of a survey on hyper-converged technology. The report noted that, of all survey respondents, "47% said they were planning to use a hyper-converged solution for infrastructure consolidation, replacing traditional compute and storage."
Your data doesn't live here anymore
And, of course, some data simply doesn't live at home anymore. Although the numbers seem overhyped sometimes, it's clear that companies are beginning to tap cloud storage more often and more deeply, freeing data center storage from unglamorous chores like storing archives. This cuts into data center storage purchases, and will likely cut more deeply as confidence in cloud storage grows.
The increased use of SaaS and cloud-based apps like Salesforce and countless others also drains data away from data center storage systems. And if you're not backing up those apps by copying the data back to the data center, your company is creating a lot of new data that simply doesn't require any in-house accommodations.
Those are just a few examples of new data storage technologies taking over from traditional storage products. And you can add software-defined storage, object, converged and probably another half dozen or so others to this growing list.
Sometimes simple is complex
The irony is that while many of the new data storage technologies and architectures might be simpler to deploy, use and manage, they may inadvertently be adding complexity to the overall storage picture. And because a lot of the newer alternatives are designed to address specific issues or workloads, you could very well end up increasing the number of siloed apps and data as you deploy more purpose-built systems.
Backup will also require some rethinking -- particularly if your shop is wedded to "classic" backup apps. Backing up data is likely going to be a lot harder with this new diversity of storage systems and architectures, which may persuade you to consider using new data protection techniques and tools, such as flat backups, cloud-to-cloud backup and erasure coding.
About the author:
Rich Castagna is TechTarget's VP of Editorial.
Old data storage techs become new again
Where is the data storage market headed?
Going green with storage techs saves green
Dig Deeper on Data center storage
Predictive storage analytics, AI deliver smarter storage
A look at hyper-convergence vs. traditional server deployments
Three trends causing the rapid commoditization of AFA storage
AFA storage tackles big data analytics challenges