 From Austin, Texas, it's theCUBE, covering OpenStack Summit 2016, brought to you by the OpenStack Foundation and headline sponsors Red Hat and Cisco. Now here are your hosts, Stu Miniman and Brian Gracely. Welcome back to theCUBE. We're at OpenStack Summit 2016 in Austin. I'm Stu Miniman joined with Brian Gracely and happy to have back on the program Balaji, Siva Subramanian and first time on the program Vishy Jakka, both gentlemen here with Cisco. Been talking a couple times with Cisco, talking about partnerships with Red Hat, of course all the updates from Metaplot and NFV Solutions, UCS, a whole lot of stuff. It's got a broad portfolio. But Vish, since your first time on the program, introduce yourself a little, what your role is and what you're doing here at OpenStack. Absolutely, yeah. So Vishy Jakka, part of the product management team in the UCS group. My focus is on the cloud solutions, specifically focusing on OpenStack. So we have a bunch of initiatives and solutions that we would like to talk with you guys. Excellent. All right, and Balaji, how's OpenStack fit into what you're working on these days? So yeah, I mean I managed sort of the cloud native solutions at Cisco. So one of the solutions is OpenStack Solutions that we do with partners. So what do you call a self-managed OpenStack Solutions? Cisco obviously is providing managed service to the metapod and to service provider focus in the NFV use case. But so we are responsible for the self-managed piece. And the other piece that I work on is on the container space. So I lead the content of open source effort, defining policy for container workloads. Yeah, so Cisco's obviously been in the compute space for a long time, doing very, very well. Like you said, we've talked about the telco side of things, we talk about metapod. Talk about how your partners are understanding OpenStack and containers and all these open things, which is sort of new in the Cisco channel and the Cisco partner space. How is that evolving for them? Yeah, I think the partners are, you know, when we look at selling any of these solutions, you know, we obviously, we're hardware, we go sell the hardware, but we're trying to find some use cases to sell to. Because customers are, you know, if you are just selling the hardware, you are sort of the last on the list of people they will talk to, right? They're trying to solve the use case problem, right? I want to figure out how do I build the cloud and all those things, so if you're just the hardware, you know, you're sort of the last one to talk to. So by providing this kind of bundle solution that we do with Red Hat in this case of OpenStack, we are moving ahead, you know, the partners are able to move ahead in the conversation. So now they are saying, you want to do OpenStack cloud, you know, we have a solution with Cisco that we can sell to, which has OpenStack, you know, fully integrated solution. So in a way, I think the partners, it's helping the partners to be more business, more relevant to the business. Right. So OpenStack, we've been talking a bunch, we were talking earlier about networking with OpenStack, we've been talking about a lot of different things. Give us a sense of how broad does Cisco think about OpenStack in terms of not just OpenStack, but where it has to link into the rest of things that are going on. What are you guys thinking at what's top of mind sort of every day in terms of saying, how do I make this solution easier? What else do I have to include? Those types of things. Yeah, so I think we look at, so there are a lot of deployments of OpenStack in the world, I think you've seen in the keynote and everything else, but most of them are mostly custom, right? We even see that customers are trying to build the custom solutions more and more, and that works, but that's not sustainable. So our goal is to make it more mainstream as possible, right? I think even now, even with our validator design, with our validator solution behind it, people are still asking for customization. So our goal is to make it more mainstream as possible. That's sort of the... Makes sense. The goal is to provide something repeatable and reliable that customers can roll out. What we have been hearing throughout the summit is lack of enough skilled people. So if companies had enough skillset, they could all roll out custom clouds, but that's not the case. And customers are looking for companies like us, trusted partners, to deploy their OpenStack clouds. And then, other day, they want to run their business, not necessarily spend their resources on standing up infrastructure and maintain a fragile environment. They want something reliable, secure, and highly available. That's what customers are looking for, and that's what we are trying to address with our solutions. So Vishy, you've had a couple of speaking sessions. I think you've done some, I have one coming up. Give us some of the highlights of what you're talking about this week. Absolutely. The sessions is a mix of what we have been hearing so far from our customers. Like Balaji mentioned, it's about use cases. Obviously, we want to sell UCS, but we want it to be more relevant to what customers are trying to do. And while they're standing up there in private cloud, they want to have a mimic a public cloud experience within their enterprise. So with managed service or with public cloud, they lose the control and data sovereignty, reliability, and security. They want to replicate all that in-house. And this is something that we have been hearing from customers. So it's essentially sharing those learnings with the customers. Then talk to them about what are the various options that customers have. It's not like one size fits all. That's not the case with OpenStack. So it's multiple options, the various options in terms of storage, in terms of OpenStack portions, in terms of capabilities. Put it all out there in front of the customers. It's up to the customers to choose what works best for them. At the end of the day, we will be with them throughout their journey. This is right from deploying the solution, maintaining it, and providing a single point of support for them. They don't want to, we don't want customers to worry about their infrastructure. We want them to be focusing on their business. That's what gets them successful. That's how that's what gets Cisco successful in the long run. So today there's a big track around containers. Containers, according to user survey, top one, two, three things that people are interested in doing. Containers, VMs change networking. Radically change networking. Things are moving around. They become virtual interfaces. Containers change it, yet again. How does Cisco look at containers, and what are some of the things that you guys are doing to hopefully make it simpler? Yeah, so I can take that one. I think I fundamentally view that the containers use a real technology. First of all, when we looked at SDN back in the days, there's a lot of hype behind it in some ways. I think I believe containers actually solve a real problem. And so we believe that containers adoption is smaller right now, but we believe that as things move on, containers would become a significant form factor for application workload to be there. So we have to make, again, our products relevant in the marketplace. Obviously, we have to add value to the ecosystem. The way the cloud-native applications are built, they assume the infrastructure doesn't, the infrastructure is not relevant. They just try to build the intelligence of reliability and scaling and all those things already on the software layers. So where does vendors like Cisco or any of the vendors make themselves relevant there? So one of the effort we are doing around this space is a project called, open source project called CONTIV, c-o-n-t-i-v dot i-o, where what we are bringing to the market, what we are bringing to the industry essentially is that as containers are getting deployed in production, more and more containers are getting deployed in production. Basically applications are getting deployed in production. There needs to be an operational policy for the applications to run, either networking related or storage related or computer related. And that is something I'm personally driving and I think is resonating well with the customers. Yeah, we wrote a piece recently for Wikibon and we basically, kind of catchy title, Docker's the least interesting part of Docker. I've got to be able to discover other containers. I've got to schedule them, keep them up and running. Like you said, there is a lot of value that has to be provided outside of just putting something in a container and going from there. Yeah, I think it's a good. As I meant to say, Cisco, the main driver is helping our customers with successful business outcomes. So it doesn't matter if you're using containers or OpenStack or any other technology. Technology, as you said, it's a piece of the puzzle. It's combining the technologies, then put the orchestration, the support, the life cycle, all that bringing all those things together is what makes our customer successful and successful at the end, in the long run. If I may add, right, the container technology is fairly immature right now. The Docker 1.9 just introduced multi-host networking, which is like, come on, right? So it's really, really early stage in terms of that. But we have done production. A lot of vendors have done that. We know that what it takes to run a production application. I think customers really, what we're getting more serious, needs to have those same policies. Just because I'm not containers, I don't necessarily have, don't have the same regulations and same security policies and things that they need to have. So I think if you look at the industry, Docker Compose and Kubernetes Part Definition, those are still defined the application. It doesn't define how the applications are actually placed or the requirements from the infrastructure. And so we are providing some sort of a framework through the content project and also some implementation of it. It's an open source project so any vendor can join in and contribute. Thanks, Danny. So let's get back a little bit. Like we said, containers very new. People are interested in a very new. Let's get back to OpenStack a little bit. We had a chance when we were talking with Lou and Rakesh from Red Hat about the multiple ways you can consume OpenStack, right? You talked about something. Give us some sense of in the self-managed space. How are you guys making it easier? Give us, what are some of those things that you're working on? Absolutely, absolutely. So if customers would, if they have to do it on their own, it's about putting, say, Cisco infrastructure and plopping any other vendors distribution on top of it. Essentially, it's two plus two equal to four kind of scenario. So what we are trying to do is, I mean, we are code designed and engineered specific plugins around Neutron or Ironic, et cetera, to leverage the APIs, the open XML capabilities and the automation capabilities that the infrastructure offers that's available out there today. And leveraging them and bubbling them up in the OpenStack context. So we hide the complexity of the infrastructure underneath behind the plugins. At the end of the, customers can consume all the capabilities of the Cisco infrastructure within the OpenStack context. And all this is co-engineered, co-developed by our partners, Red Hat, for example. So these integrations are consumed directly through Horizon, deployed and managed and configured through OSP Director, for example. So all these integrations are also the validation, the testing, the configuration, sizing, and the best practices that we offer to the customers. It makes it easy for them to deploy it. I mean, it quickly goes from a science experiment to a production-ready environment. So it's all about providing something that's production-ready, scalable, available for them. So we heard about, I mean, a lot of service so far, 65% of the respondents are deploying OpenStack. But they are deploying OpenStack over a period of weeks, maybe months. But our effort, with the joint effort, we are trying to do it in a week or two. So that's the power we bring to the table for our customers to leverage. Actually, the demo we have on our show floor shows that's bringing from a power-on to OpenStack deployment in three hours, which is actually a little, it's really great if you can just, it's all automated. So you just do it, go for lunch or whatever, and come back, hopefully, it's up and running. We have found that, like he was talking about, even just to do a park for OpenStack, people spend days trying to make the hardware and software work together. And I think this kind of effort is going to really help people to say, I want to see the benefit of OpenStack, but I want to spend weeks or months to try to just see the benefits of OpenStack. So what you're trying to do is, let's say, within a day, bring up an OpenStack for your demo or park, whatever. So then you can say, okay, I can use it. So that's sort of what you're trying to do. Yeah, so when I think traditionally, we understand some of the day-zero challenges that we always have in getting things to work. But when I think back to my networking experience, it's those upgrades that are difficult, especially most people, I put my network in place and don't breathe on it, definitely don't upgrade the code. We don't want to touch a day. It's supposed to, you look kind of the open source world. I mean, OpenStack's releasing every six months, you'd definitely do an upgrade. You'd got to be doing batches. How's that changing? How do you guys think about that inside Cisco and bake that into the product's institutions? That aspect is not easy. You're absolutely right. There's a lot of challenges there. So what we are doing is working with the community, working with the partners ahead of time before the next release comes out, before Newton comes out. We have been engaging with partners to validate the, validate the plugins, validate the various components that make up the solution. So we remove the guesswork out of an upgrade process. So that's the ongoing journey that we will plan to make it work for the customers. So one thing about that is like, I remember talking to one customer, they have like one version of every OpenStack out there. Like, you know, one part on, you know, there's a little bit, maybe a year ago, two years ago, like Diablo, this, SX and everything else. You know, because they don't know how to upgrade from once they've deployed on something. One of the benefit of our joint solution that we're creating is to be able to do that. So if a customer buys into, so we have right now shipped a few months ago on Juno, we're already working on how do I go to the next versions and they buy the solution support. We will basically provide them guidance on how to get there and we'll handhold them as they go to the next versions. Most people are actually happy with the version that they deployed. They don't necessarily need to like move every version. Even though the community is developing every six months, I don't think people would necessarily follow every six months, right? You know, because once you're in production, you don't want to touch it to your point, even though it's maybe flexible or whatever. So what we're trying to do is figure out obviously what cadence you want to have in terms of upgrades. Maybe you know, OSP7 is Juno, do we go to the next one or do we go to the following one? So these are all definitely challenging. We need to ensure there's an upgradability path and it's fully tested, validated so the customers can automatically hopefully upgrade through some sort of tools that we provide. All right, I want to give you both the last word here. You know, we talk about 7,500 people here and one of the things I think, if you talk to most people here, they're open to things going faster, changes happening. What are some of your key takeaways from kind of the attendees here, conversations you're having that you'd want to share with all of these? So absolutely, the interest is certainly there to adopt the newer version and consume the newer features but it's a journey. So they are looking at it, it's a conscious and conservative effort but the intent is to quickly adopt newer features. So they are looking for best practices talking to other attendees, talking to other companies, talking to other customers to find out what worked for them and what they want to leverage for their internal effort. It's a mutually learning experience for all of us be it the vendors, customers, or partners. It's a learning experience for throughout. So my sort of takeaway is that people are looking at containers and now they've deployed OpenStack or trying to deploy OpenStack. How does that work together? So one of the challenges do you really need OpenStack to deploy a container stack? I think that's definitely a very valid question to ask and if you do deploy containers on OpenStack you have neutron networking and then you have container networking on top. How does it all work? So that's a challenge that people are trying to solve. I think you'll sort of shake itself out in the next six months or so. So how do you deploy containers in OpenStack? And so this is definitely, I've heard like in the last week, three, four conversations from containers having this challenging decision to make on this topic. Palaji and Vish, thank you so much for joining us. Going to be right back, getting towards the end of our coverage here of OpenStack 2016 in Austin. You're watching theCUBE.