 Live from Mountain View, California, it's The Cube at OpenStack Silicon Valley, brought to you by headline sponsor, Mirantis. Here are your hosts, John Furrier and Jeff Frick. Okay, welcome back. When we're here live in Silicon Valley for OpenStack SV, this is the event in Silicon Valley for all the action for the transformation and computing in the cloud and the enterprise service providers. This is The Cube, our flagship program. We go out to the events and expect a signal from the noise. I'm John Furrier, the founder of Silicon Angle. I'm John McCos, Jeff Frick, general manager of our Cube Silicon Valley operations. Our next guest is Jennifer Lin, VP of product management with Juniper formerly with Contrail Systems, which was acquired by Juniper. Welcome to The Cube. Thanks for having me. Great to see you and we have a little sports connection we just kind of put together. It's funny when I get dressed up, no one recognized me because I dressed like a bum at the lacrosse games, but our son's playing lacrosse together. He's masquerading as a soccer or lacrosse dad when he's actually a professional producer. I know, people don't know him in two lives. But so I want to ask you, great to see you, let's talk about Juniper. Obviously, total refresh over there right now. New management team, new blood, give us a first taste of what's going on inside Juniper because Juniper was sideways for about a year or so. I mean, there's been a lot of change. We are feeling really good. We've got a new CEO in place. Shagon Herod period joins us. Both as a former customer and he was at many years at Verizon and then at Barclays Bank. And he's really coming in with some fresh ideas about where he wants to take the company. So yeah, there's been a lot of consolidation and obviously as Contra we haven't been long in Juniper but a lot of fresh thinking. So it's an exciting time. And what are you guys doing in Juniper right now specifically around OpenStack? So the Contra acquisition happened in December 2012. Juniper is a gold member of the OpenStack Foundation and a lot of our efforts have been focused around Neutron within OpenStack. But obviously given that the network touches a lot of different things in OpenStack, we've also been very engaged in a lot of the efforts around Nova and things with storage, et cetera. So as a lot of the applications become a lot more distributed, the role of the network is interconnecting those things becomes pretty interesting. And you have an open source version of the product. So OpenContrail, we have a Patchy V2 licensed all of our code and it obviously changes the discussion from a network perspective. Traditionally, Juniper is a market leader in routing, switching and security products. With Contrail, we have a pure software overlay product that obviously plays natively with OpenStack. So I was actually close to Juniper. I was working with them in 09, helped working with ProDeep, and Junos was around at the time. Is that still around? Is that going on? Yeah, so Junos is still the core operating system. Obviously the key thing that differentiated Junos a lot from its competitors was the fact that it was based on a BSD kernel and a real-time operating system like some of the other monolithic, large networking vendor players. So, you know, it's always been driven a lot by solid software devices. Can you repeat that again? I love how you said that. Say that again. No, no, no. I'll do it again. That was so good. But the good software, loosely coupled software architectural principles, it's something that I will say Juniper understands quite well. And a lot of, you know, 80% of the engineers at Juniper are software engineers. Well, I remember Juniper really was the first North, South, East, West concept around the networking. Free network virtualization. Yes. I took the eye off the ball on network virtualization. How would you guys bring that back? What's the new update? How are you guys refreshing that area? Well, I think, yeah, network virtualization, just like server virtualization, is nothing necessarily new. If you look at IPvp ends in the wide area, we've been doing that as part of our core business for some time. And obviously it scales and it interoperates with other networking players. So network virtualization is really a means to an end as people look at things like Linux containers and Docker. You know, our network solution plays very nicely whether there's a hypervisor or not. So it really isn't about server consolidation, data center consolidation. It's really about rolling out new services quickly over a pervasive IP infrastructure. And one of the things that the networking industry has been really guilty of is getting too low level into the details and every single flow and every byte and every packet. And I think that's what we're doing with things like OpenStack is really abstracting the network at a system layer. So that essentially vendors can compete around services. Yeah, so we talked to a lot of the tech athletes out there in the networking spaces. We let the geek out into the low level like that. And, you know, falling nits at Intel to folks at other companies. And it always comes down to the networks to bottleneck. So what's going to change the game on that? Network virtualization certainly does that. Is there a unique vision that says, hey, this is the spot we got to kind of tweak or is it just limited by physics? No, and I think that's a lesson that, you know, we're taking a page out of the book of the web scale players. A couple of the early control team joined control back from places like Google and Facebook where they were building out, you know, web scale architectures for very distributed applications. So a lot of our new CEOs focus is around cloud and enabling analytics for cloud and using, you know, everything that the network has in terms of its pervasive presence as an asset in that. But being much more focused around application and user policies to drive network configurations and automation. And as you know, one of the things that the networking industry has done is basically you get the badge of honor for having the triple certification and knowing all the esoteric CLI commands. And that obviously in something like OpenStack has forced more of how do you abstract various services whether it's firewall services, load balancer services, you know, presenting a network to a specific cloud tenant so that he or she does not have to worry about PGP protocols and MPLS and all of those other things that networking folks get out. So Jennifer, I have to make a confession. Now that I know that you're a super network geek, I'm going to have such an awesome time with the lacrosse games. I'm not going to hang around with the apple guys anymore. Our sons play on the same lacrosse teams and now I'm in trouble. All I do is talk to the guys from Apple what's going on with their watch. Now I'm going to talk about, you know, NFV and yeah. No way. I become a soccer mom on the weekends. Okay, all right. So we won't geek out. I will deny not, you know, I won't even. We don't know each other, you know. But this brings up a good question, right? So Neutron has been highly debated. Some are saying, you know, certainly Randy Byer has been critical of it and others. Is that in the right direction? What is the state of Neutron and or other open stack networking suites out there that's some startups doing some things in the area? But it's a big guy's game. We had heard that earlier from Pivotal. The stack is the stack. And if you're not super funded, you really can't win that right now. It's just ungettable. So you guys are heavily funded. You're a big player. Yeah, I think there's a lot of interesting issues to solve, but you're right. I mean, many of the cloud builders have come to us and said the network is the bottleneck. A lot of the challenges that we're talked about very openly at the last open stack summit with Neutron, you know, we feel like we're proactively addressing. We won some, you know, very good cloud builder deployments where they really struggled over the last two years with some of those challenges that were sort of native in Neutron. So I think we have a different approach. It caused a little bit of a discussion at the beginning, but I think, you know, the industry is getting much more towards this notion of a distributed router all the way down to the host so that we can essentially do, you know, much more low latency, scalable, dynamic routing, and really treat services not just with fast packet forwarding, but good IP principles. You know, I kind of straddle both those worlds of network and applications in my career. So I love the DevOps perspective because born in the cloud is a great thing. If you're a straight green field app developer, Amazon's great. Yep. I mean, so you hit a certain scale point then you got to stand up your own infrastructure. But at some point DevOps and the notion infrastructure as code is a great concept. Yes. Who doesn't want that, the programmable infrastructure? Now the reality is a little different. You mentioned some of the things. So when do we get to true infrastructure as that code? Is it going to be enabled by, or the question is, what is the disruptive enabler for that trend? Well, I think if you look at a lot of the startups, they're doing this already just because they have been forced to essentially do what was traditionally three siloed functions with one DevOps admin, right? And so that sort of notion that you hear from Facebook, there's one administrator for every 10,000 servers comes a lot through automation and comes a lot in understanding how to converge compute storage and networking in higher level systems. So we don't expect that every networking engineer of the future is a deep software developer, but I think even my 10 year old son is writing little Python scripts for his gaming. So there is a certain level as we get into this next generation where Minecraft and League of Legends and other things. So it's not so esoteric in terms of being, writing small pieces of code to get it done. You give him the Python book as he learned on his own. No, no, no, there's a Coder Dojo and things like that that are really fun. The kids get up there, they play games and they get pizza and then they learn Python. So it's a new world. It really is. I know. Anything to get an edge in Minecraft if they'll do anything. Exactly. That's the triple badge of honor. Yeah, I know. Good to see the little geeks growing up in Silicon Valley. So what else is new? Give us the update on Juniper, how's the management team? Any big splashes you guys making? Yeah, I mean just recently we shared some news around MetaFabric which I think we're extending to think about next generation cloud architectures. That's hitting every aspect of our business, routing, switching, security and obviously with the overlay with Contrail. We took a little bit of a disruptive move with Contrail. We open sourced our code right off the bat as we geated and that has obviously accelerated a lot of the discussions around how does the underlay play with the overlay? What does open source mean? Open Contrail which is the open source channel for us has been an amazing route to market. We've touched a lot of technology partners and customers that we probably as traditional networking folks wouldn't have necessarily gotten to. The DevOps teams and folks, hackers kind of just trying to understand what's going on. Everyone is trying to solve their own problem and traditionally the network guys don't talk to the cloud administrators and the application teams and the DevOps guys but I think we've seen a lot of that track. Do you see a day where the overlay just becomes the fabric and no one really cares what gear they're running? Exactly, we don't make a big distinction at Juniper. It's really about building a system that works and scales and there's some aspects that should be done in physical platforms. There's other things that make a lot of sense in an overlay so what we've done is we, we are right now the only pure play networking overlay that interoperates seamlessly with the physical network that's there. Whether it's a Juniper router or a Cisco router or a TriMetro router or anything else, this notion of a control plane that interoperates is something, is one thing that the IP industry has needed to do across the wide area and that's how the internet grew organically. Well that's why the server vendors are shaking in their boots because essentially the cloud is an overlay in data center so you're talking about a network cloud someday. And it's not exactly, it really is about federation of domains that have to be by nature heterogeneous but have to interoperate and integrate and there's one layer which is really about the TCPI stack, there's about connectivity from a network perspective but this broader question of cloud interoperability, there's a lot of metadata that each cloud represents that is in the application layer and that is, you know, some of that data. And multi-cloud world is, you need a fabric. Yes. You're going to be in a multi-cloud world to get multi-tenant for a minute. Multi-cloud with multi-tenants. Exactly. You're talking about, you need intelligence, you need software, you need compilers. Exactly. You need a lot of geek stuff. Yeah, and I think it's important to not reinvent the wheel. I mean there's a lot of goodness out there that we don't want to rip and replace. You know, when IP networks took off we didn't say rip out your token ring and your Apple talk and your decnet and your SNA. What we said essentially over time those applications will change as well but this layer of abstraction that we have basically abstracts what kind of physical media it is whether it's wired or wireless or what application is running on top. And that's a key principle that we don't want to change as we make this transition in multi-tenant data centers. Well, it's a multi-cloud world with mobile driving a lot of things. The pressure is coming from the outside into the network. It used to be the network where you get set up and you're done, do some things but you limit the scope. Right. Now apps are coming in and dictating terms. Yeah, unfortunately it took us four generations of IP to get to 4G LTE which is natively IP down to the handset. But we can finally run any application over any radio network, over any carrier. And that's what's allowed mobile applications to just explode and not have too constrained environment. And I want to give Juniper credit. Pradeep really had the first vision on this in 2008. He wrote a great paper on mobility apps. He really nailed it. He can sell the iPhone right away. So the question I want to ask you with the whole net screen and the whole security background Juniper has a pretty well stocked set of products and security. Perimeterless IT is a huge deal right now. So quickly share your thoughts on where you guys are at on that and how that relates to all this network stuff. I think if you look at what Juniper's done, we also no longer believe that the answer to security is a big hunking firewall at the edge of your data center. We now have 13 services in the security portfolio. All of them have been virtualized, whether it's DDoS or antivirus or distributed firewalls, IPS services, et cetera. And that becomes a key service that we're using in the Contrail architecture. We create service templates for these various security services so that we can say between your front end web server tier and your back end database, insert firewall as a service. We hope that it's Juniper's firewall, but if it's not, that service is exposed consistently into the open stack layer. Okay, great. We have some Tim Crawford, our co-host, virtually. Tim, you're in the club. How does Juniper stand out ahead of Cisco and Arista? Interesting time ahead for Juniper. Comment? Yeah, I mean, I spent over a decade myself at Cisco, so I think Cisco has a very different approach with obviously they have the UCS servers, they bought Whiptale for storage, they've got NCME and they're doing a lot of things with the Nexus team. And I think Cisco, because of the scale and the size and the history that it has, it's taken a verticalized approach to kind of solving each component. I think as Juniper, we've taken a much more software-driven approach from the beginning in that we don't expect that we own every single component of a architecture in the data center. And we've had to think much more horizontally. So I think that's a major differentiator for us. The fabric in particular. Not just the fabric, even contrail. We see that as a loosely coupled overlay component. We are already in deployment with Arista switches or Cisco switches or Juniper switches. And that's the value of the overlay model. We have to be agnostic to the physical underlay at the same time as Juniper because we are very focused on keeping this, discipline software architecture. We can show how our solutions are better than competitor solutions across the underlay and the overlay. And we're working very hard to make sure that we live up to this expectation of standards-based interoperability across the network infrastructure. Jennifer Lynn, VP of Product Management, Juniper, the action-packed man. Jay Sree was like that too when she came on theCUBE. She was like firing away. It was like, I'm like. I was VP of Product Management for contrail. We're a little senior director for data. Okay, so senior director at Juniper but you were a VP of Product Management and you're the company. Congratulations on the acquisition. Juniper's lucky to have you. Jennifer Lynn here inside theCUBE. We'll be right back after this short break. Thanks very much.