While Wall Street has been paying attention to more established players in the solid-state storage arena that have been raising money on the public or venture markets, EMC snapped up Israeli startup XtremIO for an astounding $430 million. While the purchase was no surprise, it should have an impact on the rest of the solid-state array vendors and the storage market in general.
While my friend Greg Knierieman was shocked that a company could be worth $430 million before product even ships, I've been quite impressed by the technology in XtremIO's scale-out, all-flash storage system. Sure, EMC has the engineering talent to build its own all-flash system--heck, it already has an all-flash version of the VNX and have been promising an all-flash VMAX--but to come out with a system for the 500,000 IOPS and sub-500-microsecond market that XtremIO is targeting would take EMC at least a year or two. Add in the political infighting as the VNX and VMAX groups argue that they already have a solution, and you can see why the EMC folks wanted something their thousands of sales personnel in Armani suits and Gucci loafers could sell ASAP.
Frankly, the all-flash versions of existing storage arrays remind me of my college roommate's car, a Datsun 240z with a 350-cubic-inch Chevy V8 squeezed into its engine compartment. Sure, it went in a straight line really fast, but it was so nose heavy that on turns the rear end always wanted to go around the turn before the front end.
The XtremIO acquisition reinforces my opinion that getting the most from flash requires a different storage system architecture than getting the best out of spinning disks. A flash-based system needs some sort of log-based data layout to minimize and consolidate writes while spreading write I/O across all the flash in the array, while disk-based systems need to keep data sequential to minimize head motion.
Other big storage players like NetApp (which was rumored by Israeli business site Globes to be sniffing around XtremIO during the past few months), HP, Dell and HDS are probably shopping for their own all-solid state arrays, which should make things interesting for players in the all-solid-state market, like Whiptail, Nimbus Data, SolidFire and Pure Storage, as well as high-performance players like Kaminario.
Curiously, Texas Memory Systems, the initial leader in SSDs back when they were rack-mounted boxes of RAM, has been shopping itself for a few months, with no clear takers. I'm also a bit concerned about Violin Memory. Its recent $80 million funding round put its valuation around $800 million, which might make the company too expensive to buy.
Disclaimer: EMC, NetApp, HP, Solidfire and Dell are, or have recently been, clients of DeepStorage LLC.