 Well, we're gonna get started Thanks for being up early. This is clearly all the coffee drinkers. We know where all the beer drinkers are. They're still in bed We're gonna walk through a bit of what we are doing together as Canonical and Juniper And it was designed to address something that we all are dealing with the reality of hybrid cloud So I'm John Zanos. I'm with Canonical the company behind you Ubuntu Jennifer Lynn I lead product management with the contra team at Juniper Networks So maybe we'll start by talking a little bit about hybrid cloud. I think we are all familiar with it but at the end of the day The reality is we recognize portability of applications across private and public clouds is the reality of the world we live in we recognize that The private cloud of choice that's developing is open stack and the network component of that becomes Extremely important when you think about a hybrid cloud and being able to have a note in the In the private cloud and note in the public cloud and moving workloads and viewing them consistently Yeah, and I think what we found is that as you know, folks here obviously recognize the benefits of a cloud environment Customers don't want to have to choose either public or private to that choice and flexibility They want the security and control that's typically associated with a private environment and the flexibility and the sort of on-demand And the services that are often available in the public cloud environment And a lot of the dynamic these days is that there are a lot of folks that you know unbeknownst Let's say to the administrators of a private environment are starting to put workloads in a public environment We're going to talk through some of the key use cases of hybrid cloud But I think it's that balance that makes it kind of inevitable that we will have a mix There's no there's not going to be one winner on You know the the environment from a network perspective our job is to interconnect securely those environments and make it seamless To a lot of the users or the cloud tenants that we have a heterogeneous environment underneath Yeah, and our objective is to make the process of deploying an open stack cloud in this hybrid environment and also deploying and managing the Applications that will sit on top of it very simple very automated And in essence making it very easy for a customer to look at both environments almost from a single pane of glass I think there's also been a lot of sort of quote cloud washing So I think we wanted to make the point of you know, what is what is hybrid cloud not? There's been a lot of services that have been you know traditionally virtualized and you know called cloud I think this self-service provisioning aspect of it and the X as a service model is definitely something that obviously in an open Stack environment what we definitely focus on and is the reason why open stack is is gaining a lot of ground people want that You know vendor agnostic environment where services are exposed as a service through clean and well-behaved APIs Which wasn't the case in many legacy virtualization environments And we recognize, you know people have these Hosted environments that they want to then bring in a private cloud Some people are starting with the public cloud and then moving to private clouds Some people are starting with private clouds and extending workloads to public clouds when they have spikes You know, so it's a very mixed environment and you shouldn't as Jennifer pointed out think of hybrid cloud as a particular Single use case is actually a number of use cases So a little bit about the challenges, you know, we've talked about workload portability I think that's one of the most important ones the vision that I think we all recognize Hybrid cloud cloud in general whatever we end up ultimately calling it brings to the table This is concept of workload portability. You can develop on one cloud a private or public move it to another one You're going to have applications look at both environments to put some part maybe the database on one side And the analytical package on it another it creates a lot of flexibility only if the network is able to support it Certainly security compliance all the things that I know we talk about all the time about with open stack And then there's a fundamental business challenge, you know, how does the company bring that forward? Who owns the budget? How do you manage that? And how do you get a company to think about things in? This sort of hybrid world and then it brings quite a few of Operational challenges at the very core we think the operational challenge. That's most important is how do you automate? How do you make it to reuse? So how do you deploy something? Deploy the network have it work with open stack and that not be an incredibly painful process And for those of you that dealt with neutron There's plenty of room for improvement in that area, and that's what we've tried to do with this joint solution So on the right there is just a little bit of a snapshot, you know our guest speaker today Their company pure one had done some surveys very recently Polling users around their expectations around hybrid cloud and one statistic that will Focus in on is that 78% said that in the next couple of years they intend to be using hybrid cloud services That that statistic today is less than 30% that are actually using hybrid cloud services But the expectation is that that will triple over the next three years So a lot of the market data today is needing to anticipate that this sort of flexibility between the public and private cloud environments is just a baseline capability and is not sort of a nice to have from a network perspective our job is to Interconnect securely heterogeneous environments And while a lot of the quote cloud environments haven't been built with the expectation of workload portability and sort of dynamic resource allocation Based on policies that may be cost. They may be time of day. They may be application policies You know we need to kind of bake that in up front in order to be ready for this tripling of hybrid cloud adoption I think at the end of the day, you know We view the likelihood that there will be a point in the not too distant future That everybody will be running off a hybrid cloud and they'll be doing that because they want to get into a world where they can write an application once deploy it anywhere and It would be that easy So just a quick, you know in terms of juniper contrail open contrail what we've been doing and there been some you know Very good deep dives around the approach that we've taken from a technology perspective, which we won't go into today But we fundamentally believe that the sort of federation of different domains with a unified control plane is what differentiates Contrail from many of its competitors In terms of the the heritage of IP VPNs, you know This has been the way that most enterprises connect their mission critical traffic across a shared IP win That's often outsourced to a carrier network And that has really held the test of time with the evolution of you know more dynamic workloads Etc. Contrail is really based on some standards-based work L3 VPN N systems that essentially takes that paradigm and stretches it into a virtualized data center environment so that we're not essentially Recreating how we establish secure multi-tenancy and issues around access control and how we horizontally scale private networks So that's I think been one of the areas where it's allowed us to present an overlay type of topology Into a multi-vendor physical network infrastructure and scale very quickly without reinventing a lot of the paradigms on how networks work The other major thing is obviously it's Been released to the open-source community And there's sort of a knowledge in the existing networking industry about both BGP control planes and how that works as well as how we do security between federated domains And so once again, I think that the You know not reinventing the wheel using protocols that are well tested and have to stood the test of time so that as we get into different types of Encapsulations things like the XLAN are very you know Important in the enterprise environment. We can add those without essentially reinventing the overall architectural approach This is just a snapshot. I mean if you look at a lot of the web skill folks like Facebook You know, they've been talking more about how they're creating this new data center environment where traditionally they started with you know larger clusters increasingly they're moving to smaller Pods that are sort of interconnected and making sure that they don't have over subscription, you know in the Physical infrastructure, but that they have any to any IP access across highly distributed workloads and very dynamic applications You know that they're being public about the fact that the way they're achieving that is using BGP as a control plane within the data center Obviously BGP is you know very pervasive in the wide area network But you know the web scale guys saw that pulling that type of control plane protocol into the into the data center environment Was the only way that they were going to be able to scale and that allows them to do a lot of the automation You know this statistic of managing one Managing 10,000 servers with one administrator. They can only do that with high levels of automation And so for folks that haven't seen it There's a very good blog post around how the Facebook architecture has evolved and how they achieve a Lot of the automation that they do in there in their environments, right? So this sets up an environment where is very extensible, right? The network becomes very flat and in essence, you know The vision of SDN where you've kind of abstract the network from any physical appliance and are able to make all the Nose visible as IP addresses is a key and then ultimately it allows it's important that you're able to encapsulate and move Applications around and again the importance of automation and the investments you make on that making sure you're able to reuse it time And time again, you know what we see time and time again is how do we help the marketplace? Deliver this sort of model with OpenStack as the construct for you know So to speak the cloud platform without having all the customization and unique development that happens within a Facebook or a Google for example, I think you know the other thing A lot of attention has been placed on sort of compute as a service and storage as a service Amazon obviously had a infrastructure as a service offer with EC2 a lot of their enterprise customers Wanted essentially the virtual private cloud model where essentially they could bring their own private IP address space and attach their specific network and security policies to a share to a pooled compute and storage environment so one of the fastest, you know growing services for them moves them into this private Paradigm with VPC if you look at the user experience from a contrails perspective We're really starting with that private paradigm And the workflow is very similar and a lot of the primitives that you see in the AWS VPC APIs have been mapped directly into open Contrail the contrail team had submitted into OpenStack the AWS VPC blueprint in order to essentially establish those mappings and many of our customers on one key use case that John showed before Our startups who start in AWS and as they grow they may build OpenStack private clouds on premise They want to be able to use the exact same scripts between the AWS VPC environment and their on premise OpenStack cloud Environment and that workload portability where they don't have to rewrite and reconfigure You know their their policies is very important to them So we're working with several social networking and gaming Companies where that is the paradigm that's quite different from let's say someone who starts in an enterprise financial Environment and they have this phenomenon of shadow IT where some of their application teams who really need a cloud Environment go into the public cloud and then they have to repatriate those loads back into the private environment from a network perspective Obviously, this is about interconnectivity but getting that sort of consistency between the application layer as well as you know Abstracting the infrastructure is what makes all of this hybrid cloud difficult All right, and you have to accept in this world clearly the network becomes more and more important as part of the overall solution set for a Hybrid cloud and give me that we're an open-stack You can have that conversation with addressing the reality of neutron You know, I think there's fundamentally two camps going on and the dialogue one is how do you make neutron better? Or how do you use neutron or do you use neutron as a set of APIs? What we liked in terms of what we've done together is we're clearly in the camp that says let's treat neutron as a set of API's There's a number of issues as most of you are familiar with and kind of articulated about In the slide in terms of scalability and challenges, but if you treat the SDN in this case Open-con trail as the SDN that plugs into neutron the SDN addresses these issues and makes the network quite extensible that the other view we have is that fundamentally we are constantly as Canonical looking at a cross-section of SDN's working with many many of them actually Almost all of them and we run them through our open-stack interoperability level lab to make sure we have a degree of confidence That they work with our distribution of open-stack when we start this journey with con trail We saw them as furthest along And addressing the shortcomings of neutron and creating a network as we were just articulating that's able to supply In essence the foundation for a hybrid cloud model that doesn't mean that everything's perfect But I think the yardstick for success right now is are you doing it better than anybody else? And are we satisfying the vast majority of the needs to deliver a hybrid cloud model? And that's why we've woven the solution together to make some decisions to be able to solve that very problem. I Think what we found with some of our initial contrail deployments is There were compromises in the network You know this idea of keeping sort of failure domains very small and being able to scale horizontally Were some of the issues the ability to tie in sort of virtualized services That maybe you know multi-vendor services that are presented in service templates was the other major thing that I think differentiated You know contrail from some some of the capabilities in default upstream Neutron so we because we've been working with a lot of the you know tier one carriers a lot of the large enterprises this notion of sort Of network as a service without compromise Is is a pretty high bar and you know I think that's the bar that we hold ourselves to So at the end of the day, this is a architectural diagram of how we're trying to make this work, you know Clearly we're combining a couple elements that reside in both our companies one is The open contrail SDN map through neutron to you've been to open stack as Fundamentally the open stack that's combined and in the solution We're using a tool set that we've created within canonical and tools that have been developed within juniper That fundamentally do a couple things that are really important We're automating the install so the install of open stack as part of this overall solution with the end the SDN is a very easy Exercise what we have found time and time again It's just the deployment of open stack for those of you that have gone through the process is difficult, right? So by automating that we allow getting to the functionality for that to happen much quicker the the other part is we're automating the scalability of this right we're trying to make the fact that you want to deploy a workload on an open Stack private cloud or move it to a public cloud is really not a painful process and again It's not perfect at the moment, but again, we think we're further along than any other element Certainly analytics Diagnostics and we're using our tool set juju which is a service model So you ultimately encapsulate every application You're able to weave them together so a web service with a database and you're able to automate that deployment and the scaling So it's not just a deployment exercise, but a management scale up and scale down exercise Yeah, I think this partnership has been really important to us because obviously in open stack where you're converging the compute storage and network functions We've introduced a a kernel module with the V router which is where you know We can do distributed routing and initially there was a lot of concern about well if you introduce a kernel module You know, how do we support that? You know as you know Sort of many of the kernel modules that are not upstream create a question of well, you know How does a how does a commercial support model work part of the partnership that we've that we've crafted here? We've been working very closely with the canonical engineering team to make sure that as these things are tested as a full stack the converged Solution comes together in a way that is commercially supportable That's what we refer to as contrail cloud and as as customers adopt a lot of the tools You know, there are many ways to to provision Stack the canonical team obviously has been investing a lot with their tools in addressing those You know initial provisioning as well as ongoing operational challenges at the same time You know, we've we've seen a lot of our customers contributing a bat back their best practices using things like puppet and chef and ansible Which are purely configuration management and don't address a lot of the you know ongoing automation and operational challenges that are at the system level So that's an ongoing journey. I think We make a reference here to contrail cloud because that's the sort of introduction that we did of a Ubuntu open stack plus contrail networking stack And then obviously as we've worked with the canonical team around Charming up the contrail install all of that comes together and that's something that you'll see out on the floor as well Yeah, one of the things we've tried to do with juju is make that app encapsulation easy So, you know when we talk about chef or we talk about pop it or we talk about different languages like go or bash You're able to write the charm in whatever language you're familiar with so Even the scripting that you've done with chef and puppet is preservable in the charm and reusable and what we really want to highlight Is that reusable piece right because those investments are investments that any company have already made and they don't want to lose and have to Reduce so we've tried to make the model very extensible building on what's already been done by any individual company Lastly, we recognize that Finding and managing the nodes and deploying them is really important That's where we use mass our metal as a service, which really just identifies the nose deploys the operating system Ubuntu or any other operating system and it allows for the scale up scale down model Yeah, so I think you know hybrid cloud as I mentioned before has been difficult There's a lot of questions around how do we achieve application consistency across heterogeneous environments I think a lot of the focus on containerization and this promise of right once run anywhere is Obviously indicative of the challenges that folks have had where they've had to port different application components to different versions of the Linux Kernel different storage drivers, etc That at the application layer is getting a lot of attention because it has been a very big pain point at the network layer A lot of what we've been doing with a an IP control plane is Federating heterogeneous environments if you think of the early days of IP networking The problem that folks had was that they had a number of different proprietary land protocols SNA apple talk debt neck, etc And they needed one abstraction one mediation layer for the network where these things could interconnect in a hybrid cloud scenario You now have AWS VPC and an open stack on premise cloud and VMware and other public cloud offerings And our job as a network provider is to not discriminate and interconnect those in a secure and resilient way So that network layer in between the application and the hardware infrastructure is obviously the core competence And in doing so we've extended what's worked in the past around IP VPNs and MPLS VPNs We're able to take a shared IP fabric and carve out private tenant networks and attach policies to those private networks and Obviously this discussion wouldn't be complete without recognizing that deployment can be in both virtual machines with hypervisors and You know KVM ex ex etc or with containers, you know Docker LXC For those of you that are aware as cannot we also announced this week something we're doing called Lex D Which is a light visor for managing containers. So we're taking This and extending it not only into virtual machines, but into containers as well Six months ago in Paris we announced this partnership and one of the inquiries we both got his companies was is this just another logo swap Because there's plenty of slideware and plenty announcement that occur at every open stack summit We were quite explicit that that's exactly what is to not we've been doing a number of exercises including deep engineering Engagement and improvement of our models together And Between now and then so in a six month period we announced the partnership We've done some engineering and also we've actually been out and working with a number of customers Because we think it's important when you make these sort of announcements that they are more than just an announcement of people with an intent To work together so today we're also going to have pier one talk of it Which is one of our early joint customers Why did you introduce? Yeah, so we'd like to bring up Gary He's a senior architect with pure one and code you code data services to talk a little about the journey They've been on both from a network security evolution perspective and this extension into open stack. So thanks for being here, Gary Thanks very much. I Appreciate you guys asking us to come up and talk about what we've seen happen with I Give them to the next one. You have to just do that. They're down All right. Well as a bit of context, I'd like to sort of run through the 32nd who appear one as that maybe informs a lot of our experience and why Contrail and Ubuntu has been helpful to us in making progress Peer one is a global hosting business We have a very broad spectrum of products. I think it's fair to say if If your needs boil down to servers in a data center, we can probably help you in some way We cover really all the bases between managed clouds and just space and power We've been around since 1999 So that is an age in the hosting business thing. It's fair to say and One of the consequences of that is we think we've seen a lot of trends over time We've seen a few things which didn't work We've got a lot of experience in deployment of new paradigms so Moving to a more integrated more converged cloud environment has been Something which has taken us some time as we've looked at all the options. We've been keen to do it Properly and not just make it a very siloed product is Something I think we've seen with a lot of early cloud products They've just sat very much on their own and they've not integrated with providers other products It's you can have that but it's a separate thing in a separate portal to everything else they're offering We are part of the Koji co-group of companies Our sister company is Koji co-data services who we are currently integrating our products with a bit more More than the standard marketing slide This is an interesting one because Koji co-data services operate metro networks in the Toronto and Montreal areas and We're quite excited about the opportunity to build sort of end-to-end solutions Where we can offer metro access into hosted services? and that's one of the ways in which contrails helping us as it's giving us a network paradigm which Fits in with the rest of our network I should also touch on who am I? As the guy said, I'm Gary McKenzie. I'm senior architect for P1 and Koji co-data services One of my responsibilities primary responsibility right now is for running cloud proof of concept activity. So my job is to Design what our next-generation cloud infrastructure is going to look like underneath? Talk a little bit about our network Probably unusually for a hosting provider. We have a global network. It's a full IP MP last backbone We have network presence in six countries. We are at every major internet exchange There is really We are committed to a multi-vendor environment in our network Which I think is an important thing to say it's Very easy to assume when we're talking about open-contrail, but We come to it as a Juniper customer. We are a Juniper customer We're a big Juniper customer, but we are not universally a Juniper customer We are very committed to the idea that we should have two options and each of our core network spaces We don't want to be tied into a hardware vendor. We have a concern about that So one of the things we look for when we're evaluating products in all areas is that there's an interoperability option Example being we use Juniper's MPLS view to do our MPLS traffic planning and such And that's something that works across the board with Juniper equipment Cisco brocade And that's something that's critical to us and in that respect contrail is a Good option as it's nice to see it has a proven track record with at the core level Juniper MX and Cisco ASR 9k So that's That's something which is very important to us Also on the slides just to fill it out really I put on our internet map Which we did a few years ago. This is there is an updated version now with Which is iPhone and Android app? I like to include it on these sort of things. I think it speaks well to our interest and commitment to the network Anyone who's ever been accosted by P1 sales team will probably know that we're very proud of our network and we like to sell on the quality a lot One thing Which comes out of that is obviously operating an international backbone is not a cheap exercise And we recover some of that by the fact that we can sell a network quality and our customers acknowledge that our network quality is very high But we're also very keen that we leverage what we have and We provide end-to-end services over the network that we can integrate Everything from quite deep in our cloud environments and map that on to services right down to customer premises by MPLS in a seamless manner It's one of those things which is often possible, but you end up with your engineers tying things together with string on an individual case-by-case basis and that's not something which is Scalable or supportable one day that breaks in some way and no one quite knows how it was put together So the ability to build seamless end-to-end services through the network is important to us The challenges We see from Cloud platforms are quite varied. I've split them out and I was thinking when I was looking at the slide earlier That's actually they're all technical challenges in some way But Some are driven by different things I guess is the easiest way to put it our customer requirements are I Guess the central thing we have to worry about We are a Managed hosting company in many respects our Value proposition is that we're offering a high level of service a high level of flexibility and customizability Because of that we have We have to have a great deal of flexibility and how we provide solutions. It's not acceptable to have We can do this in this way and that's it. You can take it or leave it. That's I would say the AWS way perhaps It's not Something which works for our customers customers want to they have an idea of how they want things delivered and They want to see us help them out with that So that's one of the requirements. We have a great deal of flexibility in how we deliver solutions We want the underlying infrastructure to enable us to do as many different things as possible that we can solve the problem One way is not necessarily enough if that's not the way the customer wants to solve it It's obviously it's a service provider specific challenge But it's one which occupies a lot of our time I've spoken a bit about the ability to Integrate services into a single platform. Obviously we get a lot of customers asking Why can't I manage my physical machines in the same portal as I can manage my virtuals? Why is the cloud platform over there? Why do I have a separate portal for storage? A big demand is that Everything it has a consistent management interface and a consistent level of management Which feeds into the amount of self-service which Uses a demanding that's it's obviously a feedback from the larger sort of hyperscale cloud providers AWS GCC as you or whoever you want to name That everything is self-service the thing we come across is customers want that level of self-service at times But they also want the level of support which goes with the managed product So that's something we're having to integrate and with a wider product suite with multiple ways to do things Self-service is a greater challenge than when you have one methodology The technical challenges we have are the simple ones The stack scale we deal with as a hosting provider makes Even relatively simple things pretty complex we The network paradigms are the easy one and they're I guess what we're talking about most here 4096 VLANs doesn't cut it and these stacks for us. We need a different option for that and Moving to VX LAN moving to MPLS based services is Something that needs to happen That's one of the technical challenges is we need to find a way to scale out the networks where We may have it easily conceivable we'd have 10,000 layer two segments in a larger environment Depending obviously it feeds back into the customer flexibility. There's lots of customers who they might only have you know 2030 instances, but they might also somehow manage to have 20 to 30 layer two segments linking them in various ways and That's their decision to make and we need to provide the infrastructure which enables that The other network paradigm, which is a problem is security John and Jennifer are very spoken to this that Security was very easy when your server plugged into a port and it was in a VLAN behind a firewall But when it's a cloud instance and you don't really know where it's going to be the Need to apply a ubiquitous security policy to a service hasn't diminished, but it's gone an awful lot harder to ensure you're doing it From our point of view particularly in some of the verticals we serve some of our customers. It's not just about That's we can do it, but it's about we can prove we can do it. It's about the auditability of that The last one is Something that feeds into again what we're talking about with contrail the era of having one hosting provider is over is something we've Heard a lot off from our customers. We did some research last year When we're looking at new cloud products and one of those things we got out of it is for the organizations Which are planning to deploy cloud two-thirds of them are intend to work with two or more providers or already are This is I think it's important to say when we say two or more we're not talking about their internal IT We're talking about two external providers. It's internal and p1 and aws and That means we need to build Interoperability we need to be able to connect to other environments. It's it's not optional bald gardens and not Not something that's an option anymore We have a large customer who depends on us to do the heavy lifting for their application We do some GPU based compute for them. It's something of a speciality of pure ones and They like that solution, but they also want their web front end to reside on aws for scalability for cost of nothing else and That's a it's a perfectly logical design design decision for their environment, but That's a customer. We've had for a couple of years When we built that solution, it's not pretty building something where you've got secure and consistent communication between aws Into our environments It was done in the era Prior to aws is VPC product VPC has changed things a bit VPC It's great problem is that it follows the aws methodology off There's one way of doing things. That's the paradigm it exists in if that doesn't work for you then there isn't an option but it very much Points to the fact that we need to find ways to interoperate and for different cloud providers different options whether you know It's on-premise society for enterprises as well. There isn't a single way We need to enable a lot of different ways to interoperate whether that's leth3vpn services evpn ipsecvpn from You know virtual srx using NFV on your open stack platform the ability to spin up that service and just say I want to connect this this service to something else And do that seamlessly is really important to customers and it's only going to become more so I Fully expect that that two-thirds of organizations will become four-fifths in a year to two years How appear one transitioning to open stack well One of the reasons that it's taken us a long time to look at open stack is open stack has been hard to deploy Traditionally, it's always been possible, but it's not been about For us. It's not been about that. It's possible. It's about that. It's scalable simple and reproducible now Contrails solved all the network problems and working with canonical as well I was solved a lot of the deployment and automation problems and it's allowed us to bring more data out of The open-stack environments from contrail the northbound APIs for analytics are something that looks really promising Something I'm very keen to get into soon but that information Coming out of it in a sort of systematic and consistent way is really useful also when it comes to deployments things like juju charms are a major step forward for us we feel it's It's something which makes deployment a lot more Consistent and simple to do I'll hand back to these guys Thanks Gary. Thanks very much for for sharing the thoughts there We're just gonna bring it home with just a little bit of an example in the slide that John showed earlier where many of the emerging Customers do have this expectation where not only may they write some of their early development in the public cloud and then repatriate that load into Their private cloud environment very often and we're working with one very large gaming and entertainment company where they have different tiers of their Three-tier application where the front-end web server tier because it's very spiky will run in the public cloud That gives them access to you know capacity on demand They keep their private data customer data gaming statistics in their private cloud environment now that the the issue there They get the benefit of the public cloud and the auto elastic scaling for spiky loads And this is a gaming provider that has millions of concurrent users using their platform at the same time And they don't have to move their data until they're ready to replicate it into another environment So the example here, and I'm not going to go through a lot of detail We do have a hybrid cloud scenario posted on YouTube and on the open-contrail channel We've done this environment with AWS VPC as well as other public clouds like soft layer Some of those videos are posted You know the ability to create a private cloud environment in OpenStack Name your virtual networks and create a service template in our case with a VSRX or another virtualized firewall capability That that abstracts essentially or separates the policy from the actual Configuration of the device or the service itself But exposes essentially in this case firewall as a service to the to the private and then we show essentially a workflow Where through the AWS VPC? We essentially create a virtual private cloud in Amazon insert essentially the IP address and subnet of the Private cloud environment on the corporate network and show that essentially we've established a VPN tunnel between those two clouds That solves not only the network problem and because we have a layer 3 control plane And we can carve out private subnets and have that connect with your corporate network as an extension of the private environment Obviously solves the network security problem, but it also solves this issue that people have about how do I pass my compliance and keep essentially? You know the same policies and the same scripts that I've you know tested over time And then essentially use the benefits of the public cloud I don't have to essentially go and renumber my IP addresses all of the DNS DHCP issues that often come up as you move a workload Are solved on the fly so please take a look at that we're showing some of those Demos in our in our booth as well as online just to kind of bring it home This is not something that's you know future out there We've been working with a lot of emerging customers that really need that flexibility and have already in their development processes Done that within an application different components sitting in different clouds. Yeah, so they keep in mind Maybe just to wrap this up, you know since Paris in six months I think we've accomplished a lot and as Gary pointed out we're actually doing it in the marketplace So think of this as a couple key takeaways. This is all about not only automation, but simplification and reuse It's also absolutely about using neutron as an API and then plugging an SDN like open-contrail Which can be completely successful in that doing what you need to do with the network making it extensible Make it work in a hybrid cloud that the nuance here is that none of this works If you're unable to make the deployment of open-stack and the SDN simple easy and automated That's what allows for application portability and all the use cases. We've talked about for a hybrid cloud model since I know we're pushing right up against the The time we had allocated will wrap it up and the three of us will be available outside If you have any questions because I don't know if we have any time for questions right now and I'm told no But I'll take one question Anybody you can one question? Okay. No, we'll just go out and feel free to talk to us. All right. Thank you