 Live 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, SiliconANGLE Media's flagship program. We go out to all the shows, help extract the signal from the noise. Happy to be here for the fourth year. Big ecosystem, lots of technologies being discussed here, happy to have on the program. For, I believe this is the first time, Omri Gazit, who's the vice president of products and services with Hewlett Packard Enterprises' Helian Group. Thank you for joining us. Thank you, delighted to be here. All right, so Omri, we were talking. You're actually Seattle-based, or as we like to call it, Cloud City. Can you just give us a little bit of background on yourself? What brought you to HP, now HPE? And what is your role inside the Helian Group? Absolutely. In fact, it was Cloud City in spades when I left. It was raining cats and dogs in Seattle. But I've been involved with Cloud for a little bit now. Back at Microsoft, I was involved very early on with Azure. And coming to HPE was really all about exploring Cloud from a very different perspective, from the open source perspective, and really focusing on building an open source ecosystem around OpenStack and later on on Cloud Foundry to kind of field the open source alternatives to the proprietary solutions from some of the vendors out there. Yeah, so of course we know HPE's legacy with Open. I'm curious, did you work on open source stuff when you were back at Microsoft? At Microsoft, back when I was there, it was a little bit harder to work on open source, because it wasn't really kind of the thing to do. Of course, Microsoft's really changed. Yeah, we actually had a speaker on earlier today from Microsoft talking about what they're doing with OpenStack and everything. All right, to talk to you as a little bit more about your involvement with Helion. I mean, we've had for a number of years, I remember Sargalite two years ago, it was like a billion dollars invested, thousand people working on OpenStack. Bring us up to speed as to where Helion is, how many people you have involved, customer traction, adoption, and the like. You know, as with all these things, you kind of overestimate what you can get done in a year and you underestimate what you can get done in five years. And OpenStack is no different. When OpenStack started, probably going on almost six years ago, it was kind of a twinkle in people's eye. The idea was really to try to match up with what AWS had at the time. And we found that OpenStack, probably two or three years ago, there's a trough of disillusionment as enterprises try to adopt OpenStack and really treat it like a Linux distro. And one of the things they forgot was that you can't just buy the software or get the software from upstream without learning how to operate it. And we found that enterprises get a lot more successful with OpenStack once they work with a vendor to really help them kind of transform OpenStack from just a piece of software into a solution. Whether that be buying an appliance or working with a vendor's professional services or even consuming OpenStack with a vendor operating it for them. Those are the much more successful modes of consuming OpenStack that we found so far. Yeah, so during the keynote, one of the early slides said, look, there's really kind of four big use case quadrants, Private Cloud being one of them, HPE's got a huge install base in the enterprise, rich history of helping large enterprise customers. What are some of the early use cases that you're seeing customers deploying OpenStack? What types of applications, what types of problems are they solving that you're seeing more generally deployed these days? Oh, sure. In 2013, when I first got involved, it was kind of the grizzly Havana timeframe and the sweet spot really was running cattle kinds of workloads on top of the platform. That was really the only thing that OpenStack cared about. And in 2014, it became start becoming vogue to run cloud native platforms on top of OpenStack. So things like Cloud Foundry or OpenShift and there's a little bit of controversy, which approach was the right approach, much hand wringing ensued. In 2015, we actually saw more and more pets that were starting to get run on OpenStack. And two years ago, if a customer came up to you as an OpenStack enthusiast and said, I'm trying to run a pet on top of your thing, what do you think? We would tell them as a community, you're doing it wrong. You need to rewrite everything to cloud native. These days, not so much, right? We basically have enough maturity, like life migration and those kinds of features that really kind of have pushed the platform to support that workload too. And in the last year, we've really seen NFV become a very core scenario for us. As we work with a lot of telcos, a lot of them are very excited about running both their VNFs, virtualized network functions, as well as just more traditional IT on top of the same platform. So we've really seen that as kind of an evolution and that's a great thing because now we have a platform that can support a diverse set of workloads and that's really kind of the hallmark of a platform success. If it can support a lot of workloads, then you know it's successful. So Omri, I think last year, we've general consensus from kind of the user survey and talk to people in the community is the base features were pretty much stable at this point. The last user survey, the average deployment had like 15 different projects that they're using. So I'm wondering if you can help us understand, what's new in Mataka? What are kind of some of the big things that came out and how's HPE been involved in those? Sure. Well the nice thing about OpenStack these days is that there's not just like a brand new project that falls from the sky every day anymore. A lot of the focus on OpenStack is about maturing the set of things that we have today. So things like live migration, for example, were introduced in earlier versions, but you have to kind of evolve them over versions because there's a set of features that aren't really there. So live migration, for example, has really kind of come into its own. Some of the NFV functionality like DPDK and SRIOV kind of support, which are very fancy words, but really it's all about getting to the metal and being able to talk to the network and device drivers very, very quickly without a whole bunch of intermediate layers. One of the things I love the most actually is the software load balancer in Neutron. It's called Octavia. That project has added active passive support, so now it's an HA software-based load balancer. Now we've had load balancing in Neutron for a little while now, but it's not until now that we actually have an HA commercial grade software-based solution to augment the hardware-based solutions that we have. Heat has a bunch of new functionality in it. There's just so many things that have actually kind of come into their own. Manasca's added a set of improvements. There's a lot of great things. Yeah, one of the, for a while, there was always this sort of bifurcation of are we trying to be, is OpenStack trying to be better than some of the public cloud services? Is it trying to sort of replace a VMware? You've been talking a bunch about live migration and HM, those are kind of terms people think of when they think of VMware environments. How comfortable would you be today if a customer came to you and said, hey, we're honestly actively thinking about replacing a large VMware environment. Can you do that with OpenStack today? Do you feel, is that something you recommend to customers? You know, two years ago we were hesitant. Today it happens every day. Customers come to us and they're serious about taking the plunge with OpenStack. And some of the most important parts of the maturation of the platform are really around life cycle management. So it used to be that you deploy OpenStack, so you had, let's say, a Juno release. Then you'd want to upgrade a kilo. It's like, that's hard. So you run a kilo side by side with Juno and likewise with Liberty. I would say that with our commercial product right now, we're starting to see really great levels of success upgrading people from version to version, which is hugely important. Because again, you know that the platform is mature when you don't have to leave upgrade as an exercise to the reader. So life cycle management of the platform itself has really come a long way, at least in our commercial product. The other thing I'd say is that from an operability perspective, OpenStack is really matured. We used to think of operability almost as an afterthought, but with operations being now a first-class persona, both with OpenStack upstream and in the commercial products, especially ours that we're building, things like Monasca, for example, as a scalable event store, being able to build standing queries using an alarm engine, and then have prescribed remediations for patterns that where we know things go wrong, based on our wealth of operational experience with OpenStack, running thousands of nodes and clusters, we can actually drive the right remediations to happen for those kinds of patterns that we know are common. And if you drive life cycle management with those remediations, you can actually tell your cloud to reconfigure itself based on some of the event stream that you're getting. So OpenStack, in terms of operability, has gotten really a lot more mature. Yeah. Omri, so one of the things we've been poking at, and we've talked about this with HP in the past though, is how OpenStack fits into kind of the multi-cloud world. So how do you guys look at it for not only OpenStack, but public clouds, tying into Azure, AWS, and kind of management of applications in their various environments? Well, back in the day, our strategy was that we wanted to build OpenStack-based clouds, private clouds, managed clouds, public clouds. And probably about a year, a year and a half ago, it was very clear to us that the public cloud race was a hyperscale race, and not really well suited to HP's business model. We're not kind of a race to the bottom kind of company. And so we chose to focus our efforts on continuing to double down and invest deeply in private and managed cloud based on OpenStack, and then partner with a hyperscale public cloud providers. So for example, our Paz offers, which are based on Cloud Foundry and Docker, run everywhere. They run in Azure, they run in AWS, they run on OpenStack, they run on vSphere. And so by having a deep investment in OpenStack-based private and managed cloud, but then at the same time having our development platform or cloud native application platform run everywhere, we really kind of get the best of both worlds. Yeah, so I'm curious, so you've got that kind of developer centric, the cloud foundry based stuff with Docker and everything. If you talk to most of those people, and you say, well, where do you want to live? And it's like, well, it really doesn't matter from the application standpoint, but then you've got the part of the business that does care. So can you talk a little bit about that dynamic and how you work with customers on it? Absolutely, that's exactly right. The first thing you have to do with a cloud native platform is you have to capture the hearts of developers, right? You have to give them a completely friction-free self-service experience for building and deploying applications. And the kind of sophisticated capabilities that you want to give them are things like application versioning and rollback if something goes wrong and zero downtime deployment and auto scaling and log aggregation. And there's just a wide variety of things that they're actually very interested by and they really love the kind of the Heroku style of just deploy some code, right? And sometimes they want to package everything up as a Docker container but have the same kind of capabilities I just described available to them. From an IT perspective, they want to know that they can run that same platform anyway, right? So that developers are not tied into AWS API set or Azure API set. They want to be able to do log aggregation but for very different reasons. They have PCI compliance to worry about. So they need those logs to all drain to one place. They want to do things like cluster level patching. So let's say you have the next heartbeat that comes out. They want to be able to patch an entire cluster and have all the applications on that cluster get patched at the same time instead of having to do it application by application by application. So those are the kinds of scenarios that really appeal to IT. And so you really need a cloud management platform that appeals to both of these constituencies in order to have something that will work. Yeah, so the past platforms have gone through just like you said with OpenStack, they've sort of gone through hype, trophidish illusionment, they're working out how to work in foundational types of areas. What do you see in from the marketplace? How much reality in terms of like live deployments, production deployments are things that are Docker or on Cloud Foundry versus people that are still kicking the tires? What ending are we in in terms of past, use of baseball analogy, if you will? Sure, so I would say with OpenStack, it's kind of been around the longest and it's certainly has emerged from the trophidish illusionment. We actually see a fair amount of uptake now. With Cloud Foundry, just one click stop behind that, I would say, Cloud Foundry probably last year was like people were like, well, you don't need it anymore, right? Because you have Docker and you have Kubernetes, we find at least with our customers that there's a strong interest in it as a prescriptive system that allows our customers to fall into the pit of success when they get into the world of cloud native development. And some of the rest of the technologies are a little bit behind the curve, right? So, for example, we see a lot of interest in Docker Swarm, we see a lot of interest in Mesos, see a lot of interest in Kubernetes. And with those stacks, things are still evolving very quickly. And so, customers that are visionary and are kind of looking to be really ahead of the curve are starting to adopt those technologies. Customers that are a little bit more conservative, they typically want to see how things will shake out before they make massive investments in those technologies. One of the things that they like about working with us is that we typically take a lot of the guesswork out for them. We give them the platforms that we believe are mature and we integrate a lot of the innovation from the community. For example, we took Docker, the Docker runtime, and made it part of our Cloud Foundry distribution well before it was voked to do that. We did that probably a year and a half ago. And about probably six months ago and we launched the Healing on Development Platform 2.0, which was a version two of our Cloud Foundry product, we actually brought the ability to bring your own Docker container, your own Docker image to the platform and get the same kind of capabilities that you get from the CF build pack workflow, which was super valuable to customers because they're like, okay, I can bet on this platform but I can still actually use the Docker workflow to get what I, you know, to package up my application. Right, yeah, take advantage of the hype and the developer sort of, you know, developers love Docker, ops needs something a little more structured and kind of bring those two things together. That's exactly right. That's been our strategy all along is try to bring some of the best of the open source ecosystems together into something that is enterprise grade and that, you know, we can stand behind and that IT likes and developers love. All right, so Omri, it's the last piece. Can you give us, you know, message you want to make sure customers understand, you know, users you've talked to, you know, what would they tell their peers if they were here in regards to OpenStack? So since we're at the OpenStack conference, I'll focus my message on OpenStack and that is OpenStack is ready. You know, three or four years ago, it wasn't ready. When we kind of took the journey, we've learned a lot of things and I would say that OpenStack now is ready for a wide variety of workloads. It's certainly ready to run cloud native platforms. It's ready to be able to build cattle-based applications, you know, like the cattle-centric applications on top of it. It's also ready to run pets and finally it's becoming ready to run workloads like NFV that are low latency, high performance kinds of workloads. And again, that's a sign of the maturation of the platform. The other main takeaway I would give them is that if they want to go build cloud native apps, if they want to build microservices, it's better to bet on a cloud native platform and run it on top of something like OpenStack then try to go reinvent a lot of the heavy lifting that platforms like Cloud Foundry already do for you. So that is the other main lesson that we've learned over the last couple of years is that, you know, customers that really bet on a cloud native platform get there faster than customers that are trying to, you know, have to go reinvent and run all, and learn all those lessons themselves. Makes sense, makes sense. Omri Ghazid, Hewlett Packard Enterprises, Hewlett Group, really appreciate you joining us and we'll be back to wrap up day one of our coverage of OpenStack Summit 2016 here in Austin, Texas.