So what's the latest from Pure Storage in terms of its hybrid cloud activities?
We've got a number of different areas there. One is Object Engine that, in addition to doing rapid restore by having flash on-prem for your latest backups, also allows for archival into the cloud. And again, it's consolidated data, because once that data's in the cloud, it can be used for other things, again, such as for analytics. We have the Cloud Block Store currently in beta, which provides customers exactly the same data interfaces and services that they get on-prem, which not only means that they can migrate their applications much more easily from on-prem to cloud, it means they can have hybrid applications or disaster recovery much more easily in the cloud. But it also means that, even if they want to develop solely in the cloud, [they can move data] onto other clouds or back in-house, and therefore be in a much better position to negotiate pricing. And then finally we have CloudSnap that allows customers to back up their block data base into the cloud.
Is that all related to Pure Storage's DirectFlash Fabric as well?
That's something else, although you can put DirectFlash Fabric in the consolidated data category as well. It removes direct-attach storage. Most compute architectures will put servers in a rack and tie them all together. But even if they're tied to a SAN, the servers still have disks or SSDs inside for program and scratchpad storage. With DirectFlash Fabric, they can get the same performance, but now not have DAS in the servers. So it's not only much more consolidated, much denser, but now the servers are stateless. That's a big deal. If the servers are stateless, they can be retargeted in seconds. So it really is a hyperscale architecture now applied to the enterprise or to the cloud or to SaaS companies who build their own infrastructures.
That sounds like hyper-converged infrastructure.
This goes back to the DirectFlash Fabric. If you think about it, it is disassociated storage and compute allowing the compute to be entirely stateless, meaning that it can be retargeted in seconds. With a little bit of orchestration in the software, it's basically a hyperscale, hyper-converged environment.
Could that mean that Pure Storage might some day come out with the compute part of the equation to make its own hyper-converged infrastructure offering?
No, no, no. There are lots of people who provide very inexpensive compute. So we're glad to work with them.
So you're not going to do like one of your competitors NetApp did with its HCI line?
Not by selling compute.
