The flash storage market looked a lot different at the end of 2012 than it did in the beginning. Startups continued to pocket big funding rounds, large vendors made strategic acquisitions to expand their product lines, and technology developments increased performance while bringing down cost.
You can expect another year of great change in 2013. Look for major vendors to have all-flash arrays while expanding server-side flash products, more innovation from startups, larger capacity drives and a significant shrinking of the number of flash companies due to acquisitions.
Here are some of the solid-state storage developments you'll see in 2013:
EMC should have its "Project X" array out around mid-year, based on its 2012 XtremIO acquisition. IBM is already selling Texas Memory Systems arrays; Hewlett-Packard is pushing 3PAR as an all-flash alternative; and NetApp, Hitachi Data Systems and Dell are working on getting into the all-flash market. But companies that can't wait for a 100%-flash system already have a lot of choices, with Kaminario, Nimbus Data, Pure Storage, Skyera, Tegile Systems, Violin Memory and Whiptail all shipping production or beta units.
And almost any mainstream storage array today is available as a hybrid configuration that mixes SSDs with hard drives.
Last week, Fusion-io brought out an ioScale PCIe card for cloud, high-performance computing and large Web companies. Fusion ioScale cards provide up to 3.2 TB of capacity at $3.89 per gigabyte. Fusion-io is the early market leader, but it has plenty of competition, with LSI, Intel, BitMicro, STEC, OCZ and Virident Systems all selling PCIe flash.
EMC is promising "Project Thunder," a shared storage system using PCIe flash to go with its single-server VFCache product, launched in 2012. QLogic also has Mt. Rainier coming with similar technology. Earlier this week, Violin Memory acquired GridIron Systems, whose TurboCharger appliance sits between the server and storage to accelerate data, for flash cache SAN acceleration.
The big development in 2012 was that multi-level cell (MLC) drives became the norm, replacing the more reliable and more expensive single-level cell (SLC) drives for most applications. Look for higher capacity drives in 2013. Several smaller SSD vendors have already bumped up capacities past a terabyte, and you can expect others to follow.
This has been an area of consolidation with Fusion-io, SanDisk, Intel and OCZ buying startups since late 2011. In 2013, those companies, along with remaining startups Proximal Data and VeloBit, will slug it out, as will EMC's VFCache and NetApp's Flash Accel.
SSD acquisitions in 2012 included EMC-XtremIO, IBM-Texas Memory Systems, SanDisk-FlashSoft, OCZ-Sanrad and NetApp-CacheIQ, and you can expect another long list in 2013. It will be interesting to see what happens to the smaller flash storage companies that don't get bought.
There are far more flash storage startups than can ultimately survive, or even get acquired. If every major storage vendor acquired one all-flash startup, that would still leave several flash startups on their own. And EMC and IBM have already made acquisitions while HDS is building its own flash products. One or two startups might survive independently, but some may not make it.
And it is not only startups that face survival challenges. While the industry is down to two enterprise hard drive vendors, the likes of STEC, OCZ, Smart Modular Technologies, SanDisk, Micron, Samsung and Intel all sell SSDs. Intel, BitMicro, Marvell, LSI/SandForce and Crocus make SSD controllers. Many of these are large companies that won't go away, but could exit or sell off their SSD technology.
Jim Handy, SSD analyst for Objective Analysis, said all-flash array startups will have a tough time surviving, because enterprise storage customers like the security of buying from big vendors. But he predicts that most of the startups with good technology will get acquired, even if that means the major vendors will buy multiple startups.
"Let's say one of these array vendors has significant IP [intellectual property] in in-flight deduplication, and another has a better solution for compression. There's no reason EMC wouldn't buy both of them," Handy said. "With the arrays, smaller companies are adding value, so I think it would be rare to find one of these companies that has a following that doesn't get bought up.
"It's different for people who don't add a lot of value, like client SSD makers. A lot of client SSD makers just slap components on a board and try to be the cheapest seller. Those guys will be casualties of the next downturn."