It's Jay Kidd's job to look ahead at all times. Kidd leads the Chief Technology Office at NetApp, the world's No. 2-ranked independent storage provider, with responsibility for overall technology direction, architecture, advanced development, technology evangelism and technology standards.
Each year, Kidd—who's been in the data storage business for more than two decades and has seen numerous trends come and go—publishes his industry predictions about what's coming next. Last year, for example, he correctly pointed out that hybrid clouds would become the dominant strategy in enterprise because of the increased control within them to manage transient workloads. His second prediction was that flash arrays would finally see increased growth in large and medium-size enterprise markets, and that, too, came to pass.
Kidd revealed his thoughts about 2015 storage trends to eWEEK recently, and we excerpt them here.
1) The Internet of things and big data analytics will come together in 2015 and produce "corporeal children."
"There are two 'mythical beasts' out there right now: the Internet of things and big data analytics. A lot of this shows great promise, but it hasn't quite panned out the way people expected. The IoT has been a big buzzword for the past year; everything is a 'thing,' everything is connected, your alarm clock will talk to your coffee maker and your refrigerator, and so on. Frankly most people don't give a crap about most examples people are using for IoT," Kidd told eWEEK.
"Likewise in analytics, there's been a lot of talk around what big data can do to transform customer experience, but most customers have struggled to really get an impact out of analyzing the data they have. With the IoT growing, enterprises will have access to data streams that they may not have access to before. They will correlate those with transactions streams or business process streams. Then I think the insights are really going to flow from that.
"By saying that adding these two 'mythical beasts' together, when you add the two [IoT and big data analytics] together, they will produce something meaningful."
2) The future of the data center is not about all-flash.
"Some say that the future of the data center is all flash, and that is utter nonsense," Kidd said. "IDC just published a report saying that by the end of the decade, 80 percent of enterprise data will still be stored on disks, and flash will be mainly used for the hot, active data. But there is a lot of data that is not hot at any moment in time.
"The companies that advocate a flash-only environment will find themselves niched. The companies that can integrate high-performance flash with high-capacity, cost-effective disk environments in a seamless, transparent way—that's going to become the workhorse of data management in IT."
3) The only hybrid clouds that will matter to users will be the ones where users maintain control and choice of vendors.
"2015 will be the year of multivendor hybrid clouds," Kidd said. "No customer wants to become captive to a cloud supplier. They spend their entire life in IT avoiding becoming dependent on a single supplier of any piece of infrastructure—they have multiple server suppliers, multiple storage suppliers.
"It's irrational to think that they will throw out those years of learning and become captive to a single cloud supplier."
4) Software-defined storage's greatest advantage will be when it can bridge the on-premises and cloud environments—the public and private cloud.
"This gives users a greater path to control: the ability to move data back and forth and a great sense that the cloud is an extension of their IT environment," Kidd said. "It's a discretely different silo."
5) Dockers will replace hypervisors as the container of choice for scale-out applications.
"The maturity and thought about how to build microservices applications is growing by leaps and bounds," Kidd said. "People have figured out that if I'm going to run a half-dozen Web servers with instances of business logic with load balancers, there's no reason I need a full-blown operating system to run just a single application. I can run that in a container, as long as I can isolate it from other things, and have multiple instances of it. The economics of Dockers on compute will make it a much more attractive platform.
"Hypervisors will still be important for the generic application with a bunch of different services. It's a convenient way to move enterprise applications into the virtual world. But when you're designing from scratch, I think the VM is going to be a little too heavyweight."