 Live from Boston, Massachusetts, it's theCUBE, covering OpenStack Summit 2017, brought to you by the OpenStack Foundation, Red Hat, and additional ecosystem support. Welcome back, I'm Stu Miniman joined by John Troyer. This is SiliconANGLE Media's production of theCUBE at OpenStack Summit, we're the worldwide leader in tech coverage, live tech coverage. Happy to welcome back to the program someone we've had on so many times, we can't keep track. He is the creator of the term, pet versus cattle, he is one of the OG of the cloud group. Randy, you know, wrote about everything before most of it was done, it's so good to see you, thank you for joining us. Thanks for having me. All right, so Randy, coming into the show, we felt that it was a bit of resetting expectations, people not understanding where infrastructure's going, the whole hybrid, multi-cloud world, so I mean, you've told us all how it's going to go, so where are we today, what have people been getting wrong, what's your take coming into this week and what you've seen? Well, I've said it before, which is that the public clouds have done more than just deliver compute storage and networking on demand, what they've really done is they've built these massive development organizations that are very sophisticated, that are, you know, really come from that web-scale background and move at a velocity that's really different than anything we've seen before, and I think the hope in the early days of OpenStack was that we would achieve a similar kind of velocity and momentum, but I think the reality is is that it just hasn't really materialized, that while there are a lot of projects and there are a lot of contributors, the coordination between them is very poor and, you know, it's just not the architectural oversight that we really needed, isn't there? A couple years ago at the OpenStack Silicon Valley gave a presentation called The Lie of the Benevolent Dictator, and I charted a course for how we could actually have more of a technical architecture oversight and just that really fell on deaf ears, and so we continue to do the same thing and expect different results, and I just, that's a little disappointing for me. So what is your view of hybrid cloud? You know, no disagreement, you look at what the public cloud companies, especially the Big Three, the development that they can do, you know, Amazon, 1,000, you know, new features a year, Google, what they can do with data, Microsoft has a whole lots of applications and communities around them. You know, we're mostly talking about private cloud here. It was a term that you fought against for many years. You know, we've had great debates on it, so how does that hybrid play out? Because customers, they're keeping on premises, Edge fits into a lot of this too, so there's not one winner, it's not a zero sum game, but how does that hybrid cloud work? You know? Yeah, so I didn't fight against private cloud, I qualified it. I said if it's going to be a private cloud, it's got to be built and look and smell the way that the public cloud was. Right, if it's just VMware with VMs on demand, that's not a private cloud, that was my position. And then in terms of hybrid cloud, I don't think we're there yet. I've presented on this at many different OpenStacks. You can see it in the past and I sort of laid out what need to happen and that didn't happen, but I think there's hope. And I think the hope comes in the form of Kubernetes and to a certain degree, Helm. And the reason that Kubernetes with Helm is very powerful is that Kubernetes gives us a compute abstraction so that you don't care if you're on the public cloud or OpenStack or VMware or whatever. And then what Helm gives us are charts. So ways to deploy services, not just software. And so what we could think about doing in the future is building a hybrid cloud based off of Kubernetes and Helm. Yeah, so Randy, since last time we talked, you've got a new role, you know, Juniper. Juniper had done a contract acquisition, you know, quite a few years back. You wrote a good blueprint on one of the Juniper forums about the OpenContrail community. So tell us a little bit about your role, your goals in that community. So OpenContrail has been primarily a Juniper initiative and we're gonna press the reset button on the OpenContrail community. I'm gonna do it tonight and call for people to sort of get involved in doing that reset. And when I say reset, I mean, wipe the operating system, reload it from scratch and do it really as a community, not just as a Juniper run initiative. And so people inside Juniper are very excited about this and what we're trying to do is that we believe that the path forward for OpenContrail is ubiquitous adoption. So rather than playing for just the pieces that we have, which we've done a great job of, we want to take the world's best SDN controller, we want to make sure everybody uses it because we think the aggregate that's good for not only the entire community but also Juniper. So love the idea of kind of rebooting the community in the open, right? Because you have to be transparent about these sort of things. That's right. What are the community segments that you would like to see join you here within the OpenContrail? What kind of users, what kind of companies would you like to see come into the tent? Well, anybody's welcome, but we want to start with all of our key stakeholders that exist today. So first one and arguably one of the most important is our competitors, right? So we're hoping that you have Rantus at the table, maybe Ericsson, Huawei, anybody, Cisco, hey, come join the party. Second is that we have done really well in SaaS and in gaming. And we like to see all of those companies come to the table as well, work days and after that and so on. The third segment is enterprises. We've done well in financial services. We think that that's a really important segment because they're a leading edge of enterprises typically. And then the fourth is the carriers. Obviously incredibly important for Juniper. Folks like ATT, Dorsetelicom, all those companies we'd love to see come to the table. And then that's really the primary focus. And then anybody else who wants to show up, anybody who wants to develop in Contrail in the future, we'd love to have them. Well, with open source communities, right? There's always a balance of the contributors and developers versus operators. And we can use the word contributors in a lot of roles. Some open source community is much more developer focused. Others, more operator focused. Where do you see this open Contrail community starting out? So where it's been historically is more of our end users and operators. I think that's interesting and an interesting twist because I think sometimes open source communities get stuck with just the people who can contribute code. I'm from an operator community myself, so I think that's really interesting. We still want all those people, but I think what has happened is that when people have come in and they want to be more sort of on the developer side, the community hasn't been friendly to them. And so that's a key thing that we want to change. When we were talking to certain carriers, they came and they said, look, it's great, you're going to do this, we want to be a part of it. And one of the things we'd like to contribute is more advanced testing around VNFs. And I just look at that and I'm like, that's what we need, right? Juniper is not, can't carry all the water on having sophisticated test suites for VNFs and more advanced networking use cases, but the carriers are deep into this and we'd love to have them come and bring that. So not just developers, but also QA, people who want to increase the code quality, the architectural quality and the aggregate value of OpenContrail. Okay, Randy, can you help place OpenContrail where it fits in this kind of networking spectrum, especially there's open source things. We've talked about VPP a couple of times on theCUBE here. The joke for many years was SDN was still does nothing. NFV solutions have grown, have been huge use cases, really where the early money for big deployments have been for OpenStack. Where does OpenContrail fit? Where does it kind of compare and contrast against some of the other options out there? I'm going to answer that slightly differently. I've been skeptical about SDN overlays for a long time and now I am helping with one of the world's best SDN overlays. And what's changed for me is that in the last year I've seen key customers of contrails, of junipers, actually do something very interesting, right? You got an SDN overlay, it's complex, it's hard to deploy. You got to wonder, why should I do this? Well, I thought the same thing about virtualization, right? Until I figured out sort of what was the killer app. And what we've seen is a company, one of our customers and several others, but one in particular, I can talk about publicly riot games, take containers and OpenContrail and marry them so that you have an abstraction around compute and abstraction around networking so that their developers can write to that and they don't care whether that's running on top of public cloud, private cloud, or in some partners' data center globally. And in fact, they're going to talk about that today at OpenContrail days at 3.30, well, those are going to present a lot more details. And that's amazing to me because by abstracting away and disintermediating the public clouds, you actually have more power, right? You can build your own framework. And if you're using Kubernetes as a baseline, you can do a lot more on top of that computing network abstraction. You talked about OpenContrail days. Again, my first summit, I've actually been impressed by the foundation, acknowledging there's a huge landscape of open source and other technologies around there. OpenStack itself doesn't have to invent everything. Can you talk a little bit about that kind of attitude of bringing, I mean, we talked about Kubernetes and that sort of thing, but all the other, you know, CNCF projects, monitoring, even components like SCD, right? We're talking about here at this conference. So can you talk a little bit about how OpenStack can interact with the rest of the open source and cloud native, writ large community? That's sort of a tough question, John. I mean, the reason I say that is like that the origins of OpenStack are very much NIH. And there has been a very disturbing tendency to sort of reinvent the wheel. Great examples, Keystone. Still to this day, I don't know why Keystone exists and why we created a whole new authentication standard when there were dozens and dozens of battle-tested, battle-hardened protocols and bits of code that existed prior. It's great that we're getting a little bit better at that, but I still sense that the origins of the community and some of the technical leadership have resistance to organizing and working with outside components and playing nice. So it's better, but it's not great. It's not where it should be. Really, OpenStack needs to be broken down into a lot of different projects that can compete with each other and all run in parallel without having to be so tightly wound together. And it's still disappointing to me that we aren't doing that today. Randy, I wonder if you can give us a little bit of a personal reflection. You've been involved in cloud many years. We've talked about them in the state of it. Where do you think enterprises are when they think about their IT? How IT relates to business? Some of the big challenge they're facing and kind of the just rapid pace of change that's happening in our industry right now. Yeah, well, the pressure's just increased. The need to pick up speed and to move faster and to have a greater velocity, that's not going away. That seems to be like an incredible macro trend that's just going to keep driving people towards the next event. But what I see is that the tension between the infrastructure IT teams and the line of business hasn't really started to get resolved. You see a lot of enterprises backing into using DevOps as a way to try to fix the culture change problems. But it's just not happening fast enough. I have a lot of concerns that basically private cloud or private infrastructure for enterprises will just not materialize in the way it needs to for the next generation and that the line of business will continue to just keep moving to public cloud. All the while, all the money that's being reinvested into public cloud is increasing their capabilities in terms of feature sets and security capabilities and so on and I don't see the materialization of private cloud happening very well at this point in time and I don't see any trend lines and tell me it's going to change. Yeah, what recommendations do you give today to the OpenStack Foundation? I know you haven't been shy in the past about giving guidance as to the direction. What do you think needs to happen to be able to help customers along that journey that they need? I don't give any guidance to the OpenStack Foundation anymore, I'm not on the board of directors and frankly, I gave a lot of advice in the past that fell on deaf ears and people were unwilling to make the changes that were necessary I think to create success and even though I was eventually proven right, there doesn't seem to be an appetite for change and I would say that the hard partition between the board of directors and the technical committee that was created at the outset with the founding of the foundation has led to a big problem which is that there's simply business concerns that are technical concerns and there are technical concerns that are business concerns and the actual structure of the foundation does not allow that to occur because that hard partition between them. So people on board of directors can't actually tell the TC that they'd like to see certain technical changes because they're business concerns and technical committee can't tell the board directors they'd like to see business changes made because there are technical concerns around them and I think that's just that it's fundamentally broken until the bylaws are fixed. So Randy beyond what we've talked about already what's exciting you these days you look at like the serverless trend is that something that you find intriguing or maybe contrary view on it I'm curious what's exciting you these days? Serverless is really interesting in fact I'd like to see serverless at the edge I think it would be fascinating if Amazon web services could sell a serverless capability that was actually running in the mobile carriers edge so like on mobile towers or in central offices if you could do distributed computation for IoT literally at the very edge of the network that would be incredibly powerful so I am very interested in serverless in that regard. With Kubernetes I think that this is the future I think I've seen you know most of the other initiatives start to fail at this point Docker Incorporated just hasn't made the progress that you need to hopefully change the leadership will fix that but it does mean that there are more and more people are gravitating towards Kubernetes and that's a good thing because whereas OpenStack is historically got no opinion Kubernetes is a much more prescriptive model and I think that actually leads to faster innovation a greater pace of change and combined with Helm charts I think that we're going to see an ecosystem develop around Kubernetes that actually could be a counterweight to the public clouds and really be sort of cloud agnostic private public at the edge who cares? Randy Baez always appreciated your very opinionated viewpoints on everything that are happening here pleasure to catch up with you as always John and I will be back with lots more coverage here from OpenStack Summit in Boston thanks for watching theCUBE.