 Live from San Francisco, California, it's theCUBE at VMworld 2014, brought to you by VMware, Cisco, EMC, HP, and Nutanix. Now, here are your hosts, John Furrier and Dave Vellante. Okay, welcome back everyone. Live in San Francisco, this is VMworld 2014. This is theCUBE, our flagship program. I'm John Furrier, my co-host Dave Vellante, our next guest is Vikram Bambri, VP of Product Manager at EMC Advanced Software Division. Welcome to theCUBE. Thank you. At EMC we're one of the exciting things we did there was we had two queues, we had a chance to talk to a lot of folks, ASD Advanced Software Division, doing some really high-end stuff. You know, obviously EMC's new third platform as Joe Tucci talks about, very software-centric. So, give us a quick update. Obviously EMC and the Federation, part of the VMware family all together, what are you guys doing here? What's the software message for you guys? Thank you, thank you for inviting me here. So, at the broadest level, if you think about it, right, the story that VMware is talking about, the software-defined data center story, and Viper and the products that we built inside of ASD plug very nicely into that story. And the way to think about that story is, there's Viper which provides the software-defined storage layer both from a control plane perspective but also from overall providing newer services like object, HDFS, block, et cetera, on commodity hardware, so which fits in very, very nicely with the overall software-defined story. And specifically in context of VMware, we have actually now with the VCloud Air, Viper object is the underlying technology that's powering the object system in VCloud Air. So that's another pretty powerful, one additional feather in the cap and it's all part of the Federation. So it helps the story really well. So why the choice of Viper for VCloud Air? I mean, other than the fact that you guys are part of the Federation and lots of love there, but there's got to be at the end of the day some justification from a technology solution standpoint. Yeah, now I think at the very fundamental level, we've built Viper to be the next generation of cloud storage systems. So it's been written from ground up with core cloud principles in mind. It's been written with like no notion of downtime, geo-replication, redundancy, erasure encoding for lower storage footprint, et cetera. And this is not the first time around that we are building this product. Having done this in a public cloud environment before with some of the engineers who are now working on ASD, having built Atmos before. So this is truly our third generation of the cloud storage product. So Atmos, you got Azure DNA, right? Yeah, so Amitabh, Myself, Manavir, and bunch of other folks who've actually come from Microsoft and built Windows Azure before. So doing this, leveraging the same learnings from that environment, doing Atmos, and now bringing all of that technology and leverage into doing Viper platform has been really, really helpful for us. We just had Google on talking about Kubernetes, talking about the Docker announcement with VMware, and the theme that the gentleman from Google, Craig, was resonating was really about bringing sort of the Google style of application development that's just to the enterprise, sort of opening that up for the world to benefit from. And obviously they'd like to benefit commercially from that. We've seen a sort of half a decade or more long trend of the sort of hyperscale approach bleeding into the enterprise. And that's sort of what you guys are doing. So where are we in that transition? How fast will it occur? And how substantive is it? Is this a sea change? Is this a niche? How broad is it? It's a very interesting question. And I think fundamentally the industry is shifting towards that hyperconvergence. And you saw the announcement this morning from VMware. There are a bunch of other competitive players out there who are moving in the hyperconvergence space. And I think there's going to be a mix of both hyperconverged as well as two tier architectures. But specifically talking about Viper, from day one when we instantiated or started working on Viper, it was written from ground up with the premise that one day we actually want to enable that hyperconvergence in the platform. So if you look at talking about Docker, even before Docker was released as a GA product, we adopted that in our Viper infrastructure. So we made a big bet on Docker even before Docker was out there. And we used that container technology to deploy Viper code on the underlying hardware infrastructure. And given that Docker already enables the deployment of random applications in that workload, you can only see the next leap of faith is going to be how do we progress the Viper software to be able to deploy compute workloads alongside that as well. So you guys discovered Docker before the big mainstream. A lot of folks have, we caught it early too. Jerry Chen invested in it with VMware connection, VM Mafia, VM Mafia. He's first investment, right? Well, you know, I kind of hinted to the sophomore Jinx. Maybe he got such a good deal on his first one. How do you top that? So maybe you on one unicorn, you're good. But what did you guys find compelling about it? And what specifically were you guys doing? Obviously advanced software group guys doing some pretty cutting edge work in all of our interviews. It's always been top shelf technologists at EMC. What did you guys, how did you identify it? How did you stumble upon it? And what did you guys do with it? So again, there are a set of people in our organization who always continue to look at OpenStack and what's going on in open source technology. And we've been sort of following Docker since it's very inception. And based on the simplicity and the low overhead of delivering that container technology without taking on a much bigger burden was like, you know, we thought, okay, we'll give it a try. See if it works. And as we started to dig deeper into the implementation as we started to look at that for our specific use cases, we realized more and more that it was actually the model that we were looking for. VM model is absolutely good. But I think there's a specific set of use cases for Docker technology that we're going to showcase as we move forward. I mean, basically bottom line, App guys love this. Absolutely, yes. So, okay, App guys, it's always been the DevOps culture, the infrastructure guys. Just give me access, give me faster networks, give me more disks, I got to run my app. Right, right. There's a culture shift there, right? There is, there is. So this is a nice bridge. Absolutely, yes. Yeah, and that bridge is actually simplifying the whole story of bringing the convergence together with like, you know, now you have storage layer that can expose out multiple protocols like object, HDFS block, file over time as well. And then you tie that with the application stack on top using like, you know, the container technologies like Docker. And now, in one shot, you have a full conversion for structure running in your environments. And that's a wonderful story to tell, not only just from like, you know, a architecture perspective, but in actual implementation, it just works. This is the agile holy grail. I mean, if this plays out with the way people are kind of getting intoxicated with it today, it is standing up stuff fast, breaking it or fixing it, being agile. I mean, that's what we're talking about, right? Yeah, so one of the big advantages of the public clouds has been around being able to take advantage of that quick, like, you know, depth test model, being able to deploy a break fix and move things around. Now we're bringing all of that into the enterprise space. So that's where like, you know, that there's a good sort of match between the requirements that the enterprises have had for a while, but they haven't had the technology in order to enable this. So this brings it on together. We always saw with the swagger of DevOps people and you know, we always joke about Mark Zuckerberg. Dave and I are always in the queue, say, you know, Mark Zuckerberg at Facebook, the poster child of DevOps, certainly Google Amazon there too, but you know, Mark Zuckerberg, the young executive billionaire, move fast, break stuff was their motto. And then recently they changed it in their last developer conference to move fast and be 100% reliable, because they have a huge load on the system. So in the spirit of phase one of innovation, you're breaking stuff, but you're innovating, right? So what's breaking in the infrastructure? The pressure that this must push on storage and the infrastructure must cause a ripple effect. What do you guys see as an opportunity, and I say break stuff, I mean stuff that's not going to be broken, but like legitimately innovation. What's happening downstream from that? From the infrastructure and the storage perspective, the thing that's really important is stuff is going to, you have to go in with a premise that hardware will fail, you have to go in with the premise that like, you know, things will go bad in the data center. So if your software is designed from ground up to ensure that it can handle failures and it can recover from failures very quickly, that's the key part. So when we built Viper again from the principles who are making sure that there's no single point of failure, you always sort of create fault domains and things like that in the platform so that any time there's a failure, you're able to recover from that failure very quickly and be able to move on and keep the system running up all the time. Okay, so hardware will fail, you got to have super fast recovery with this little human intervention as possible. Absolutely. Eventually, the components of a rack become disposable. Is that right? Put it into the shredder. Absolutely, yeah, the way we think about it, everything inside, for Viper, we consider it as a fault domain. So a single disk is considered a fault domain, a node is considered a fault domain, a rack is considered a fault domain, and data center is considered a fault domain. The building itself. Yeah, and anything could fail at any point. You have to be prepared to be able to handle that failure. And if you build that resilience in the software layer, then in that case, the problems become a lot more simpler to handle, because yeah, you can afford those hardware failures, you can afford data center failures and still continue to make them work. Okay, then back to one of my original questions about the opportunity here. Why wouldn't every data center in the planet want to run like that? So why isn't this the future? I think eventually it will be. It's just that the technology is still comparatively new. People have implemented or large service providers or ISVs have implemented in their own data centers first. And now the enterprises are definitely catching up very quickly. So if I had that conversation like three years ago, people would have been like, yes, I know there's cloud, but I want to take my pace and think about it. Now we are having a proactive conversation where the enterprises are approaching and saying, I actually want to talk about cloud first. I want to talk about this commodity infrastructure first. And yes, you can talk to me about rest of the hardware that EMC produces as well, but the first point of conversation needs to be about what's happening in OpenStack, what's happening in the commodity hardware infrastructure. How can I leverage object, HDFS, these newer data types in a much more meaningful way? And the homogeneity of something like Azure based on sort of your experience or of observing things like that, whether it's Google, Microsoft, Amazon, the homogeneity of that infrastructure allows those organizations to succeed where is it the heterogeneity of the enterprise that creates more complexity and slows down adoption? Can you talk to that? Sure, I think public clouds are successful because of the fact that they're able to create that homogeneity in their infrastructure underneath, but it ends up presenting a certain set of challenges because it does not meet every single requirement from a workload perspective. Now, when you take that back into the enterprises, enterprises for the right reasons have heterogeneous hardware that exists in their environments today because of the application workloads, because of SLA requirements, et cetera. For us, it was very important to make sure that we don't take away that flexibility from the enterprises. Still give them the same features, the same functionality, the better cloud economics, but not to take away that agility or that flexibility of choosing the specific hardware. So the way the Viper platform's been designed, you can run services on top of either traditional hardware, you can run on commodity hardware that the customer brings to the bear, or you can actually buy ECS appliance from us. So again, the choice is in the hand of the customer and they can pick whatever platform works for them. So give us update in the product set. So two years ago at EMC World, you guys announced with a lot of fanfare, Viper, classic Jeremy Burton product, it was enough ready to be introduced, but it wasn't shipping yet and you guys shipped a little bit ahead of schedule actually. But so where are we today? You've announced Elasti Cloud Storage, you got Viper 2.0, give us the update on the product. Sure, so Viper controller has been out in the market since September 2013, and we actually have a large number of enterprises and service provider customers now taking on and putting into their production environments. So CSE has like 175 petabytes running Viper, SAP is running a large production environment in their environment. So lots of big adoption happening in the enterprise space. Viper on the services side got launched again in September 2013, what a truly commodity based storage platform got launched in Viper 2.0. And that's been now we are working again with service providers, enterprises, enabling private cloud for them. VCHS or VCloud Air is now like a prime example of where we're leveraging the storage system inside their data centers. And then Viper services as an appliance, that's what Elasti Cloud Storage is. It's think about it as public cloud storage in a box and that we just announced at EMC World and now it's in production and we are selling it to our customers. Pickup, do you have what I would call born in the cloud service providers or application companies using Viper? Obviously, VMware, you mentioned CSE, huge customer, obviously. But how about any born in the cloud guys using it? Yes, there are a few customers that are truly born in the cloud third platform applications. I can't talk about it for, but like, yeah. There are a set of customers who are looking at like Viper as that commodity play, enabling all of the cloud services on top of commodity hardware in their data centers or in a service provider data center. All right, we got a wrap, but give us your bumper sticker on VMworld 2014. John and I often ask that question. Truck's pulling away from your perspective, the advanced software division at EMC. What's the bumper sticker on VMworld 2014? So we actually have a very great story with VMware and that is the division combined together. Viper controller and Viper services essentially give you one stop, like, you know, both control plane management as well as a storage platform built on scratch running on top of commodity and third party hardware. And it integrates seamlessly with VMware stack. There's built-in integration with Vasa provider. There's built-in integration with VC ops. And also there's for orchestration and automation. There's also integration with VCOC. So in the end, you can have all of your complete stack running compute storage and networking in combination with Viper and VMware, right? Vic Rump, thanks for coming on the queue. Really appreciate it. Vic Rump, Bambri, VP of product manager at EMC, advanced software division. Obviously they are not sleeping at night. They are coding away and they're at the software-driven enterprise. You guys are doing a great job. And it's great to see you guys on top of Docker. I mean, people look at EMC and think, you know, spinning disk company, but really a lot of software involved. Certainly, Jeremy Burton is clear on that when we do the chats with him. Thanks for coming on the queue. Really appreciate it. We'll be right back live from San Francisco here at VMworlds2Cube, our 50 year. We'll be right back after this short break.