If you’re confused about all the action with EMC, VMware and Pivotal over the past several months, you’re not alone. CEOs have traded places, joint ventures have been struck, product lines have been sold and GE even came on board. And that’s before you even start talking about all the new technology.
I sat down with EMC SVP and CTO John Roese on Tuesday at the company’s annual EMC World conference to find out what’s up. Here’s what he had to say.
On three companies under one roof
While they’re technically three separate companies, EMC is really in control. It’s the majority shareholder in VMware and owns more than 60 percent of Pivotal, its new joint venture with VMware that includes the Greenplum, Pivotal Labs, SpringSource, Cloud Foundry and Cetas business lines. When it comes to everyone working toward a common goal, Roese said, “The good news is that while there is independence, Joe Tucci is the chairman of all these companies.”
Roese calls himself the “gravitational center” of the three companies when it comes to technology. This is a reinvention of the CTO role at EMC, which used to be more of a research position. Now, he puts the stake in the ground and generally directs everyone toward it, even if they’re not all taking the same path to get there.
On why Pivotal happened and why it matters
My takeaway from Roese’s comments on formation of Pivotal is that Greenplum is really the linchpin of the whole company. At its core, Pivotal is about building big data infrastructure that can handle next-generation workloads, but it’s aware that broad adoption is only possible if that high technology becomes easier to consume. That means new higher-level applications, which is where SpringSource, Cloud Foundry and Pivotal Labs come into play.
All of this technically could have been accomplished by just selling Greenplum and Pivotal Labs (the only assets of the new company that was under the EMC umbrella) to VMware, but Roese said VMware wasn’t the right home because VMware is not so important in the places where next-generation workloads are popping up. There’s not a lot of VMware inside carriers’ data centers, he acknowledged, but there is a lot of OpenStack popping up. And there’s a lot of Amazon Web Services everywhere you look.
“We would like the big data infrastructure to not care about that,” Roese explained. From EMC’s perspective, it doesn’t need to own the middle — the cloud operating system, if you will — if it can still engage customers at the storage and application-platform layers.
On keeping independent while working an ‘unfair advantage’
Roese doesn’t think a vertically integrated approach is the best way to do business in today’s technology world, which is why EMC, VMware and Pivotal all operate independently and no one relies on another in order to work within customers’ data centers. That’s why VMware has its own cloud computing efforts but Pivotal is cloud-agnostic, why EMC storage can operate with any higher-level software and why VMware doesn’t care about what’s running underneath or, usually, above it.
However, he added, it’s only natural the three companies seek an “unfair advantage” from the incestuous bonds they share. What he means, of course, is that they should keep a close eye on what the others are doing and work together to ensure they’re all optimized for the same types of workloads. For example, Roese said, if EMC didn’t reconsider how storage had to perform given that virtualization is the norm or that technology like Hadoop exists, it would “become suboptimal or generic.”
The same holds true for Pivotal and VMware. Pivotal needs to think about how big data applications run on virtualized resources differently than on big bare metal systems, as well as on flash-based arrays like what EMC is about to roll out based on its XtremIO acquisition. VMware and EMC need to think about how their software-defined data center and software-defined storage approaches can build off each other.
From EMC’s perspective, it’s easy to see why this all matters. It is at its core an information infrastructure company, but “the challenging thing with that is that it’s a moving target,” Roese said. A company like EMC can’t get by on storage arrays alone anymore, but it also can’t be dumb enough to think it can be everything to everyone and still be good at anything.