Despite major shakeups among its large IT-vendor peers over the past few years, storage giant EMC maintains that it’s more than capable of sticking around for the long haul and housing data for even the most-innovative types of applications. This week, Jeremy Burton, president of products at EMC, came on the Structure Show podcast to explain the company’s plans for the years to come.
It’s actually quite an interesting interview, especially if you’re interested in hybrid cloud computing, software-defined storage, and how EMC claims it can deploy those things for customers faster and cheaper than previously possible. Here, however, are a handful of highlights that help to illuminate the company’s future plans more than its current product offerings. It remains to be seen whether EMC can continue to sell its systems in a world where words like commodity are praised and legacy or even enterprise can have negative connotations, but Burton does paint a compelling picture.
Next-generation apps will run on Hadoop
“Over time here, we’ll start to allow the IT folks to not just provision traditional storage arrays under a software-defined storage layer,” Burton said, “but kind of true commodity hardware that will run underneath Hadoop instances, which is going to run I think the next generation of applications, as well.”
The company has been making sure its new products, such as the software-defined ViPR system and the stealthy DSSD technology it acquired earlier this year, will treat Hadoop’s file system as a first-class citizen for exactly this reason.
Hardware is overrated …
As the mention of commodity hardware suggests, EMC knows that highly engineered hardware and Hadoop don’t often go hand-in-hand. In fact, Burton largely downplayed the importance of hardware sales to EMC’s business at all.
“Old tag lines or, I guess, images die hard,” he said. “I have about 15,000 people who work for me, and I have 300 that do hardware.” Most of them work on design, he added.
And with cloud computing rapidly gaining acceptance among enterprise customers as a storage option, hardware will become even less important. Of course, there might be limits to the types of providers EMC is willing to work with when it comes to storing customers’ data:
… unless it’s really important
However, Burton was quick to point out that while hardware won’t be too big a deal in some cases, it will very much matter in others. EMC is betting that real-time applications and the internet of things, in particular, will drive demand for in-memory databases and storage, and those will require “super-high-performance infrastructure underneath.”
“I don’t subscribe to the view that folks are just going to roll up to their Chinese ODM, buy a bunch of disk, stuff it underneath some software-defined storage, and magically it’s going to give the performance characteristics that some of these in-memory databases are going to need,” he said.
In the next year, he added, we’ll start to hear more about the technology EMC acquired from a secretive, Andy Bechtolsheim-backed startup called DSSD, which is geared to this brave new world:
“… It natively talks HDFS to the applications, it actually doesn’t even smell like a storage array at all. It doesn’t talk iSCSIs and Fibre Channels. It talks natively what the application speaks.”[/blockquote]
It’s worth noting that DSSD isn’t EMC’s only foray into in-memory storage. The company, along with spinoff Pivotal, is also backing the Tachyon in-memory file system project. It was developed at the University of California, Berkeley, and is a close cousin to the popular Spark data-processing framework.
Big data and IoT will drive business
Of course, all of this innovation isn’t without reason. EMC is building software and infrastructure that it thinks will be critical to its ability to keep selling storage as those big data and IoT applications start taking the lion’s share of customers’ capacity.
“I think it’s going to be the big revenue stream from EMC three, four, five years from now,” Burton said, referencing the complementary set of Hadoop, ViPR and DSSD. And, he added, that Pivotal spinoff will definitely start to pay off at that point.
“Paul Maritz would be my best friend and his [cloud and big data software] would sit on top,” Burton said.