 Live from Santa Clara, California extracting the signal from the noise, it's theCUBE covering Nextwork 2015. Brought to you by Juniper Networks. Now your host, John Furrier and Stu Miniman. Okay, welcome back everyone. We are here live in Silicon Valley in Santa Clara, right in the shadows of Levi's Stadium for special Nextwork hashtag NXT work, Juniper Network's customer summit, this is theCUBE. We go out to the events and extract the signal noise. I'm John Furrier, my host Stu Miniman with wikibon.com, industry analyst in cloud. Our next guest is Jennifer Lin, senior director of product management at Juniper Networks. You're the contrail, guru, congratulations. You have been part of Juniper Networks. How many years now? It's been almost three years. Almost three years. So, CUBE alumni. Yes. Welcome back. Very happy to be. Yes, thank you. So you just came back from Tokyo. A lot of action going on in cloud. Juniper is right in the middle of all the action. Certainly don't necessarily have a cloud of their own per se, but they're an arms dealer. They sell a lot of gear to the Amazon, everyone else. But really the network is really the action is. So we see SDN, we just had the CEO on, talking about some of the cool things. But what's going on in the middleware and cloud pass is really critical. Bridging, pass and cloud, platform as a service with networking has been a work area for you for a couple of years. We've talked about it before. What's going on? Give us a quick update. So here are big announcements. Give us a quick update. Sure, yeah. The industry is definitely changing. I think a lot of people are looking to simplify how network automation and service delivery is done. A lot of focus on open source platforms like OpenStack. Contrail obviously was acquired by Juniper a couple of years ago. And I think we're starting to see a lot of maturity of the customer segments who are trying to get to this sort of simpler system level automation. And abstract a lot of the complexity that has been in networking for some time. And around cloud deployments, the convergence of network servers and storage definitely changes the discussion. So we have lots of new discussions with customers that are definitely making it more exciting. And certainly the trends are all over the world. You're seeing it, EMC being bought by Dell and the big consolidation of storage, companies, Western Digital, HD Oracle's doing some stuff. But the word glue always comes up in the software world. So the glue is critical. So you've been working on a key area that has to enable but also connect in a very integrated way. What are the key things that customers should know about? Contrail, platforms of service, some of the networking innovations that Juniper's doing. Because it's now all kind of lining up. Yes, absolutely. And there's glue in there now to use that term. You made it abstract. Yeah, I think one way is glue. Another way, I mean, our job is as a networking software platform is to interconnect and federate domains which are often very heterogeneous, right? Now that the maturity in cloud has gotten to well, there's going to be public cloud. It's not going away, whether it's AWS or Google Compute or another public cloud. And a lot of our customers are building their own private cloud environments. So we have to interconnect those. Obviously Juniper has had a leadership position in routing, switching security. And now with the overlay model, we can drive a lot more of a flexible software model in how to stitch those things together, whether it's virtualized or physical infrastructure. And that stretches not just within how do we automate data center systems but now with things like cloud CPE to the branch access domain as well. Quickly talk about the clouds to be it's a big announcement here at the event, customer events is being showcased. That was enabled by Contrail. Talk a little bit about that dynamic. Yeah, absolutely. So cloud CPE in terms of, you know, what we can do in sort of changing the discussion around the CPE devices but also how does that get managed at the system level? Contrail service orchestration, obviously, the term orchestration is really pointing to how do we templatize the delivery of these types of services? That the notion of network as a service and security as a service and load balancing as a service are things that, you know, kind of change the game. How do we get sort of vendor agnostic templates in software so that as these things happen dynamically, we don't have to reconfigure physical devices and that's a lot of the flexibility that we're, you know, really doubling down on with the software defined model. So Jennifer, you were at the OpenStack show in Tokyo last week. My team was, yes. Oh, your team was, thanks. We've seen at least Neutron is starting to mature and customers are starting to feel a little bit better at that. But one thing we look at, there's so many options. I mean, I count at least eight to 10 different, you know, major companies, startups in the space. There's got to be some consolidation and, you know, reduction of some of those options. What do you see happening in the OpenStack networking space and can you give us a little color as to where is Juniper contributing to OpenStack? Yeah, actually. So OpenStack is sort of hitting a certain level of maturity and I think the most complex part of rolling out these large scale systems has turned out to be the network. You know, by abstracting what the network behavior is supposed to be through the Neutron APIs, I think we've sort of, you know, hit that balance there's a certain amount of complexity that's needed to present those network services, whether it's routing or switching or security or load balancing services. And at the same time, there are a lot of options in terms of how to approach that. Neutron very much started out as sort of plugins to configure VLANs. And now it's gotten, you know, like with systems like Contrail, we've abstracted a lot of the intent of the network and how do we deploy network and security policies without sort of getting into very low level commands in CLI. And you know, now that we've been through a number of bake-offs and trials, I mean, we've been very successful with, you know, different segments in terms of scale and performance in HA and, you know, how do we do service chaining? You know, a lot of the discussion is now, you know, how do we get through the deployments as easily as possible and how do we make sure that we have a flexible architecture so that if it's one day virtual machines in the next eight containers, you don't have to revisit the architecture. Okay, as we look at Contrail, how are you looking to, are you looking to build really an ecosystem around it? Especially I look kind of from the northbound standpoint, you know, what services, you know, should we be seeing, you know, love to see a logo slide of, you know, all the companies that are saying, I'm committed, I want to do it, and therefore, you know, choice is there and there's a lot of options because today there's a few options out there that are all trying to, you know, kind of build that ecosystem. Yeah, I mean, we've been working very closely with a lot of partners, you know, more broadly in the OpenStack ecosystem as well as VNF partners, right? So as we enter into, you know, network function virtualization solutions, we have a lot of partners who, you know, their services are delivered either as virtual appliances or even as containers, and we can work with them as an ecosystem. We've also worked with folks like, you know, Morantis and Canonical who, you know, partner with us both at the Linux OS level, as well as the OpenStack deployment and professional services roll out. And that's been very successful because as a software component, we've been very loosely coupled so we can work with, you know, multi-vendor environments, both from the service layer as well as from the network infrastructure layer. So interoperability in terms of the network services has been key, and Juniper has made that commitment in terms of being very open and software defined. There's a very clean software architecture there that's sort of underlying both the physical routing and switching as well as the OpenStack components. Yeah, so Jennifer, you brought up containers. I wonder if you can help unpack for us a little bit what you're doing. We were at DockerCon. There's big announcement on Docker networking. I would expect to see Juniper more part of that. You guys have, you know, are your customers asking about it where is it now? Absolutely, so, you know, as far as contrails concerned, we've supported Docker and LXC containers for a while now, but I think the key change moving forward is now people want to see this workload portability. So whether it's a virtual machine on any hypervisor or it's a container, we can take that workload and move it across any cloud environment or any cluster within a data center. And we've kind of really been seeing a lot of interest in that, you know, we can not only pack more containers into each host, but we can also move containers across public environments or into the private cloud. We have one customer, Lithium, did a keynote at the OpenStack Summit. They talked a lot about how they're doing their initial development in AWS and then moving that workload into their OpenStack on-premise cloud. From a network perspective, it works the same way. And from a contra perspective, whether it's on a hypervisor or, you know, right on the Linux kernel, it works the same way. All we're doing is creating virtual networks that are secure and defined with a policy and making sure that as those workloads move around, we can still enforce the VPN policies. That's awesome. I want to just drill into that. Is how much setups involved in that? Is it all automated? I mean, sounds like magic. So we have zero-touch provisioning for it. I mean, just like cloud CPE, I think a lot of the challenge has been around avoiding a truck roll or avoiding a lot of manual, you know, installation. Now, you know, with the automation scripts, whether it's, you know, Puppet or Chef or Ansible or Saltstack or something else, folks have ways that they want to define what that behavior is meant to be. And then deploy it across a number of servers in the data center or across a number of CPE devices. So that zero-touch provisioning model in a way that as changes happen in those endpoints, that the orchestration system can handle it without filing a trouble ticket and recreating, essentially, a lot of it. So you're never aware from the sense of container aware. What do you call that? Container aware, workload aware? I mean, what's going on? What's the interplay? Can you just drill down on that? Yeah, so the problem that I think a lot of our customers are solving is that the infrastructure needs to be transparent. A lot of our customers are very focused on moving more quickly. So let's say for our SaaS providers, the reason they're building these clouds is because they want to keep their developers rolling out new revenue-generating applications without having to worry about all the infrastructure changes that need to happen. So in the old days, you know, the application development teams would have to wait six to eight weeks to get their new environment going. Now they essentially want to self-provision the environment, roll out, quickly make changes. And, you know, they're not doing two releases a year, they're doing 20 releases a year. So the agility that they need in the infrastructure, they shouldn't have to care whose vendor infrastructure it is, what services. Those are all essentially being automated at a much higher level. Like Amazon has had, you know, a lot of sort of success in their VPC offering and that's why they're growing in a lot of the enterprises as well. Tell me about the dynamic of integrated stacks and integrated apps or infrastructure. We set Oracle open where we heard we heard the engineered systems which kind of makes sense from a vertically integrated if you want to harden that, but horizontally scalable has been the ethos of open source. Yes, exactly. Horizontally scalable, but yet, are they mutually exclusive? There's probably use cases where you want to have a vertical stack at some level in the stack, but it's all going to be based on the workload. Again, creates more complexity. Yeah, I think, you know, with what we're doing, we're kind of building a horizontal platform just like the IP network in the wide area has, you know, been sort of, you know, with VPNs as an overlay on top of IP networks. You know, similarly in the cloud, I think the verticalized applications where you hard-coded the application to, let's say, Sun Solaris servers or, you know, specific hardware specs has kind of gone away, especially as you're talking about in web services or open source type software. So now, x86 servers for the most part look very similar. Obviously, the specs may change a little bit, but most folks are kind of shifting the value up into the software layer when they can and making sure that they can sort of leverage the scale and economics that a lot of the cloud computing guys are sort of driving because they're rolling out 200,000 servers at a time. So I think those verticalized systems used to have to be defined because you had Oracle ERP systems or Microsoft Exchange tied very much to a specific hardware spec, and now with open compute platforms and that kind of thing, you know, people are kind of agreeing, well, there's sort of a certain definition of a cloud computing infrastructure from an x86 Linux perspective, and on top of that, people will add software value. That's phenomenal. Well, thanks so much for coming on theCUBE and sharing your insights. Oh, thanks for having me. This is great, you're a CUBE alumni and for people who don't know, CUBE alumni, people who have been on the CUBE before, we call them CUBE alumni's, and they join the exclusive list and you can find that list on LinkedIn. It's only members only of CUBE alumni's. Jennifer, thanks for coming, sharing the insight. Always great to talk to you, want to get more time, but I know you're super busy and thanks for coming on theCUBE. I appreciate it. Thanks for having me, really appreciate it. Jennifer Lynn, sharing the data here about cloud and all the software is super complex, but the automation is the key message here at Juniper Network's customer event, customer summit. We'll be right back with more for this short break.