Twilight of the pure-play storage gods

Storage has achieved sudden celebrity status with the rise of dynamic new technologies. The explosion of choices means that IT leaders must take a close look at the new vendor landscape.


Not so long ago, the world of storage was pretty straightforward. All conversations were about external storage, and external storage was all about storage area networks (SANs) or network attached storage (NAS). It was all very predictable and manageable but, truth to tell, not too exciting.

That’s changed in a big way in the past few years, as one wave of innovation after another has swept over the sector (I recently met with blogger Calvin Zito to talk about some of the biggest storage trends – check out our video interview.)

Looking around today’s very different landscape, I’m constantly struck by a couple of new bedrock realities that have profound implications for anyone who’s charged with shaping their organization’s IT resources for the future – and for any vendors vying for that business.

1. Once you’ve tasted flash, there’s no going back. Flash is a fundamental, once-in-three-decades technology transition, just as significant as the one we went through when spinning disks overtook tape. Now that solid-state drives are overtaking spinning disk, it’s becoming clear that flash memory can deliver the performance that today’s applications demand in a much easier, more predictable way than spinning disks ever could. We’re reaching a tipping point; storage architectures are rapidly becoming all-flash architectures.

At the same time, flash is still a misunderstood phenomenon in the enterprise. When considering an investment in flash, many IT organizations simply compare the unit cost per gigabyte with the unit cost of a gigabyte of disk. I’d submit that that’s a very incomplete way of looking at it. The dirty little secret behind spinning disks is that because they’re so inefficient, storage systems have to be over-provisioned to match the kind of performance that flash delivers. So even though the unit cost of flash media may be a bit higher, at a system-cost level flash can be less expensive. And, of course, it comes with a much smaller energy footprint. When you layer in data reduction technologies like deduplication, the financials become even more compelling.

The cherry on the top is the dramatically enhanced end-user experience you get with flash. Put it all together, and it’s clear that flash will continue to breathe new life into external storage for a long time to come.

2. Software-defined is unstoppable. More and more storage conversations are starting with this question: “Do I really need external storage? Or is server-based internal storage the right architecture and the right infrastructure building block for me?” The shorthand for the latter is software-defined storage.Four strands dominate this discussion:

  • Object storage. When unstructured data explodes in scale, the traditional scale-up and single-node file systems can no longer provide the right container for it. This data is much better suited to object storage – scale-out file architectures – which in turn are better suited to server-based internal storage.
  • Shared-nothing architectures. These are massively scalable architectures in which each server owns its own disk resources. NoSQL databases and Hadoop traditionally use servers as building blocks, clustered in a shared-nothing configuration.
  • Hyper-convergence. Hyper-converged appliances provide a tight coupling between the virtualized layer, the application layer, and the storage intrinsics underlying both. A single hyper-converged box provides storage, compute and network fabric, ideally under the control of a single management tool. That adds up to easy, fast deployment and hugely simplified operations.
  • Composable infrastructure. In this new category – the next evolution of converged infrastructure – storage, compute and network fabric become fluid pools of resources under the command of a comprehensive software intelligence and a single API. (Here’s a great infographic that describes what composable infrastructure delivers.) Data services can be provisioned on the fly and changed on the fly. Storage becomes an integral part of a true infrastructure-as-code platform.

Software-defined storage is absolutely here to stay. Servers with internal storage are becoming a very credible infrastructure building block for delivering data services.

Shelter from the storage storms

Given the slew of new options, which way should a storage buyer lean? The absolute best answer is … it depends. Every business will have different needs, of course. But as IT decision-makers pore over the product catalogs, they should also spend some time reviewing where the various vendors stand in the market and how they’re positioned for the future.

The storage space has seen more action, and more disruptive innovation, in the past five years than in the previous twenty-five. And there’s no sign that the storage storms are over. One fundamental change I see is this: the glory days of the pure-play vendors are fading fast. The pure-plays are not equipped to take advantage of the fact that in many cases compute is now the right infrastructure building block for delivering storage services.

You don’t need to look much further for evidence than the recent news that a major storage player, EMC has sold itself to a much smaller (in terms of market cap) generic compute vendor called Dell. EMC has essentially thrown in the towel and admitted that they can’t go it alone. NetApp is another vendor that’s looking around for a tighter coupling with compute. Whether its partnership-of-convenience with Cisco will work out for the long haul remains to be seen. 

Of course, IT leaders can also choose from a large number of pure-play start-ups that dot the storage landscape. But it’s a brutal environment, and the mortality rate is high; historically only about one in ten storage start-ups that get funded survives. With pure-play storage itself now a declining category, the number of potential customers for the start-ups is declining too. The question that buyers have to ask themselves is whether any point-in-time technology benefit they may perceive outweighs the business risk of dealing with a company that will likely have a short lifespan.

Many IT organizations will prefer to work with a full-stack vendor that offers the complete range of storage options alongside comprehensive compute and networking portfolios. The choices in today’s storage space are wider and more exciting than ever: disk vs. flash, internal vs. external arrays, traditional vs. software-defined. Full-stack vendors can deliver the right solution – regardless of whether it’s server-based, storage-based, converged, or composable – depending on what best fits the customer’s needs.

If you'd like to know more, Vish Mulchand, Senior Director, Product Management and Marketing, HPE Storage, offers his thoughts on the pitfalls of selecting the wrong vendor in all flash storage here.