 Okay. Welcome everyone. My name is Lou Tucker. I'm a VPN CTO of Cloud Computing at Cisco Systems and thank you for coming to the session. Today what I wanted to talk about and I think we've seen a lot of this and much of this will probably be a repeat but maybe what I'm going to try to do is give you some color of my view on what we're doing here with OpenStack. But it's really looking at how OpenStack I think over the last couple of years and in the near future will continue to expand in a lot of different market opportunities, a lot of different applications. It's not just in an ordinary cloud platform anymore. We really are in many ways I view this as becoming verticalized into a couple of very interesting industries. At the same time as what we're doing is that we're increasingly aggregating a lot of new technology that's coming about even though the four or five years that OpenStack has been around. So I think actually Jonathan you know we're all you know talking about these disruptions that are happening in the industry today and these waves of innovation that are fueling companies like we saw Uber mentioned earlier and we saw you know Airbnb. These are major disruptions of industries that are happening because of these technology innovations that we've had that allowed new business models to arise and other people you know and allowing people to approach these problems in very different ways. So we're both seeing you know the entire you know who goes to a store to rent the CD anymore. It's all being delivered down by Netflix. Amazon now is becoming the largest retailer almost in the world delivering things again through you know the internet purchasing through the internet and delivering then even in the same day delivery through logisticals large logistical centers around the world. So all of this is in many ways I think you know just driven by the internet and by cloud computing and the reason why that's so is that the internet gives us the reach we can now address very very large audience audiences marketplaces and cloud computing allows us to deal with a very small number of people. So when those two things come together the internet and cloud computing you have the possibilities that very small companies can take on very large industries and go after the entire marketplace and that is something that I think we are all a part of and Mark Adresin made this point back sort of in 2011 that essentially softwares eating the world and it could eat into all of these different businesses and that's what people are experiencing today and that there's no notion that software itself because of the ability for us to rapidly create software requires a lot less capital intensive businesses we can approach these businesses in new ways because driven by software and those so all of you who are developers you've got a very very bright future because software really is taking over more and more of the traditional enterprises and so if we think about this in another way that we've already passed that that time period in which now there are more virtual machines and physical servers out there traditional IT department you'd measure the growth of IT or whatever by how many servers how many networking ports and things like that and now it's a number of virtual machines which kind of interesting about virtual machines as you know they come and they go they disappear they may be around only for a couple hours or a couple days so it's a very different model that we're talking about provisioning now resources it's on demand it's self-service none of us like to file a trouble ticket or call up somebody on the phone to get resources for us we wanted to be able to to go to an API and and type in a command and get resources provision for us and so it started actually in actually virtualization as you know goes way back to IBM mainframe days and when it started in the enterprise there was still essentially run by IT organizations so you still have the file of trouble ticket but now the IT organization's job was easier because now they could actually just spin up a couple of virtual machines for you instead of racking and stacking servers I think the big change in cloud computing was when that became self-service and that's what I think we have Amazon to thank for their model with the self-service model and everything just like FedEx no longer are you calling up to see how you track your package you given a link in an email and you click on the link and you can now track your own package well that saves FedEx an awful lot of money and that provided you with a great service so the notion of self-service combining again with the virtualization technology I think really sparked this whole change in cloud computing and then people started you know looking at it and like like Joe Wyman if you've read any of his material really looking at it from the point of view of economics and from an economics point of view you can start talking about things such as you only have a certain amount of sustained capacity you need every day and then you have peaks around holiday seasons or or anything else so the notion of hybrid cloud starts to come about where you are using resources not only that you own or that you lease for a long term but also then you through peaks and valleys here you can more accurately match the demand curve with your cost curve that directly translates into profit companies and so that's why there's a strong strong economic foundation behind everything that we're doing in cloud computing and something else happened not only was this a different way to do the same kind of computing we used to be doing that we changed infrastructure but we started looking at these cloud native applications and so this changed sort of the design paradigm for building applications you started to think immediately about the fact that on these virtual machines since you can spawn them up or or or kill them off over time you can make it dynamically elastic and so that became an integral part and when you're dynamically elastic that also means you can recover from failures if a couple of those virtual machines that are on servers that have a problem or you've got a disk failure or something it doesn't matter because you have a fleet of virtual machines that you're dealing with so a lot of the resiliency of applications started moving resiliency of the application moved into that application layer and was less dependent upon the resiliency of the underlying infrastructure platform because we know at scale those kinds of failures are just going to happen every day and you have to have a service that's running nonstop so the notion of these kind of cloud native applications application design patterns started to come about okay so now we virtualized everything we are running your own infrastructure in the cloud and I see like we have got Ryan here here who's been doing it in one of our super users look talking about the Webex platform and a lot of things he talks about are the fact that now even though you virtualize that some of the things are very much the same you still have to operate it so you are now still operating your virtual machines and so the notion of DevOps is really paramount here at two different layers from the system provider point of view where you have code driving the infrastructure and then again at the application layer where you have an operational concerns about keeping these things up and so whatever you have operations you have operational costs the way you control operational costs through automation the way you get automation again it's through software so all ties back to this is the whole notion I think that Andreessen was talking about several years ago it's interesting to look in the in the past a bit here over we've open stack because we've seen now over the years the number of different applications and industries that we're talking about that now we are seeing being addressed by open stack we've gotten this not just cloud providers most of these are not cloud providers in fact when you're providing open source software that's so anybody can run that software and so people are building all of their their private clouds within the enterprise they're running clouds to run e-commerce operations even so we see like Bloomberg and Comcast and Best Buy these are major industries we saw you know Walmart talk about their their fleet of you know over a hundred thousand cores that they're running now so this is all now very very possible it still requires quite a bit of work because this is still very new technology that we're talking about but these companies unlike Nicholas Carr's does you know statement many years ago does IT matter well for these companies IT matters a lot as a matter it's fundamental to them because they're out to disrupt other companies with IT and so that becomes very important role that we have in terms of making that making open stack continue to address these kinds of applications so in some way I think that we now are seeing open stack becoming this new layer in the data center we are looking at a point we had physical data center we sort of had administration and orchestration and operational tools usually managing that physical infrastructure and now on top of that we layer a set of open stack services and it's not just a compute service not just a compute and a storage service or compute storage in a networking service this is over time moving into database services moving into message bus services I think that over time we'll increasingly see things such as Monaskia coming up which is talking about big data pipelines for you to be able to process your logs and events and everything else so all of these become services and in fact that's a new platform those of you who follow at any of the what I've been talking about a couple of years I always objected to this definition of platform as a service yet infrastructure as a service platform as a service and software as a service I just thought the platform part we got wrong should have been services become the platform that's what I think is really happening so that when people now are building applications they're drawing upon a number of different services and the competition now is do you have those services in your cloud do you have those services available to your users because that's what makes a developer's job easier nobody wants to stand up their own database anymore if you can have one provided as a service so those are the kinds of things that accelerate the application development that that is going on you know at the same time cloud computing was coming about there's another shift happening in the industry and that's around software defined networking we've all heard about open flow and how the same model that we were talking the the model around virtualizing resources that we saw applicable in the compute space is now being applied into the networking space and so this allows the kind of perfect storm between software to find networking happening and cloud computing happening and so the now companies and some of you know the largest like carriers are looking at network function virtualization they're saying the net and what that means if you decode that NFV is really talking about those traditional network services that we all use in applications load balancers firewalls VPNs those are becoming virtualized now so that they can be dynamically spun up instead of having to make a phone call saying please provision my you know company with VPN services you can do that through self-service you can dynamically scale those services and you can reduce the operating costs for the provider which means that they should be passing those cost savings on to the consumer as well so these things are coming about at the same time and colliding with with cloud computing so what do we see again open source I think is one of the other real drivers of this of this change as well and so we're seeing even open source now moving into the world of defining you know designing these network controllers to open daylight now is really growing up and coming on very very fast where a lot of the functionality we want and all of this networking will be encompassed by these controllers and so this has a lot of the same constituents as open stack that is we're getting together and we're designing this kind of an open source network controller that we all can use and then plug into different infrastructures to provide these kinds of network work network functions if you look at new trying itself it's rather interesting we started this very more than like three or four years ago at something called quantum and ridgeline now is neutron and this was in a way to say let's make networking as a service that's actually something new we had no real model to go after so it's been through a couple of iterations and what we see here was designed very early we wanted to enable multiple technologies with SDN coming on we wanted it to be able to you know accommodate open flow we wanted it to also accommodate traditional networking if you have just got standard networking out there it should work with standard networking as well and a lot of different models so in fact there's over 40 plugins that are built into neutron today provided by a lot of the different vendors and different open source communities which allow people to essentially plug and replace the different networking components that they would like to have I think that we're going to see continue to see a lot of change in this area in fact in Europe there's a standards body called FC who is defining what they what a virtualized network service network function virtualization should look like architecturally so standards bodies are getting involved here because they want to define a reference standard or architecture here that would allow multiple companies to adhere to this and plug in so that the consumer of these could plug in different components from different different vendors and so this is work that is coming out of that on the code side so that's a standard body on the code and the implementation now that is being done by open NFV and in concert with with our summit here we've also had a day with the open NFV folks as they're looking to build a reference architecture around network function virtualization so this is this isn't something four years ago we weren't even thinking of when we're starting OpenStack that was going to be a cloud platform for IS now we're looking at it also being able to accomplish being the the platform the virtualization platform for network function virtualization entirely different space to do that even on the foundation side on the rest of the community here we've had to spin up different groups OpenStack has to now get to be carrier grade enterprise ready the resiliency of the platform itself becomes really important so there's a lot of work that is being done and I think it will continue to see going forward and I urge you to work behind a lot of the blueprints that we're bringing out to really make OpenStack resilient have the same kind of uptime that you'd expect to have in a carrier space or in the enterprise space the other day I had a talk actually with Bank of America it's interesting even the banks are getting involved in this most traditional part of IT and what they're looking for they are very much behind this notion of a software-defined data center software-defined infrastructure they want to think of their entire IT infrastructure as being driven by software that means they can automate it to reduce the cost that means they can be more agile they can roll out new services to their branches as the bank and to new branches that are showing up in shopping malls or pop-up stores around the world they want to rapidly provision new services and they can only do that through software lowering they have to be very cost conscious so it has to be fully automated and they have to de-risk their investment and one principle way of de-risking their investment is to go to an open-source platform such as OpenStack where we have they can get a solution for multiple vendors and they don't tie necessarily their new services or their applications to a particular vendor and be held hostage they like the fact that it's multi-vendor and participation by broad aspects of the industry so in many ways you know configuration and cabling those of you who've ever worked in IT or worked in data centers know the nightmare that this is and that's now being replaced by code and that's what this whole software-defined infrastructure is largely about so that you can have Yang models you can have formal ways of expressing what you want the infrastructure to look like and instead of sending guys out there with a trouble ticket that they have to review and cut and paste from one system into another system of record and a CMDB they can express it now through a software model and that software model becomes the model that then the rest of the orchestration system is supposed to make the infrastructure look like and we're only looking then at the difference so that when there are failures you can say there's a failure here because what we're actually running doesn't look like the model and we can produce that kind of a feedback loop so the other notion that's somewhat is sort of the holy grail in IT in many cases you would like it to be policy you don't even want it to be this you know JSON file or Yang model or whatever you would like it to be expressed as a policy the intent of what you would like to have you would like to have the intent expressed through these mechanisms so that you can say these web servers will they have to talk to the internet but my database server is only behind doesn't have no connection to the internet so that you can talk about the policies that you would like to see enforced you would like to be able to say the front end of these servers here which is some internal service has both a firewall and has a load balancer so that any of those individual nodes could go down and the service continues to run that's intent and that's a policy so there's many efforts underway as this could have been very heavily involved with group based policy which is API's and now allow you to express groups sets of servers your web front end your middle messaging tier perhaps and your back in database those are groups and you don't talk about the individual IP addresses you talk about traffic from this set should be able to go to this set filtered by these following rules and that's the intent that you'd like to do so a lot of the mistakes that we make when we get down to translating intent into configuration or even config files is alleviated because you're instead talking about intent and that intent can apply then and wherever this application might be put inside of the infrastructure and your binding sort of with the IP addresses and everything else starts to go away there's all the other things in terms of policy Congress is another effort underway to really define even a more higher level policy across both applications and services and infrastructure and I flex then reflect policy going down into the individual devices so watch this space over the next several years is more and more applications you're going to find another level of efficiency through this kind of automation by again what as computer scientists what are we always doing we're abstracting things so we can talk about them in a way that removes the dependencies upon the underlying layers and so policy I think will be a very important aspect of that a couple of weeks ago I went to Las Vegas for the NAB show National Association of Broadcasters I don't know if you've ever happened to be in Vegas I mean this show is going on but it is huge it's like CES is one of the things that takes over a large part of the city and since this is all about media and broadcast and TV you know big screens flashy screens all over the place and the latest movies and sitcoms and TV shows are being talked about this entire industry has been built on hardware appliances you couldn't use general purpose computing you were doing transcoding video ingest you're doing broadcast where latency is paramount high bandwidth is required and so that that's just in this show itself we're seeing the set of companies that a couple years ago started looking at it going you know look at what happened in other industries and we're going to see the same thing happen in ours and it's kind of start to virtualize so these functions if you're transcoding every time that we come up with you know to get the standard broadcast out to this and after all of the other devices that's different encodings that we have to take place upon even accommodating for different bit streams and bandwidths so all that video pipelining now which has been done on hard and appliances is moving into the cloud computing realm so do we see things for example today that those applications that were in these fixed appliances are now moving into a virtualized pipelines and some of the early movers there are doing this on open stack so I find that particularly exciting because here again is a whole other industry now that's starting to move to cloud computing and so that five years from now NAB show will look very different it'll look more like a software show where people are showing the tools and everything else where much of this now is virtualized and so obviously the advantages you get you know is that you're cutting down from weeks to minutes one of the things in the broadcast space this is all real time many many years ago I worked on real time signal processing and there are real strict constraints I mean you just have to be on there you can't lose frames and broadcast so you really have new things but at the same time they have workload requirements they have 45 movies they just got in and they have to transcode and encode and get them ready by Friday so they've got huge demands in terms of the quantity of work that they're trying to achieve in a fixed period of time so the orchestration of those different components and the resource allocation and scheduling placement of those services becomes very very important so here's another industry that's being I think disrupted by cloud computing and by OpenStack and I think this one will be a very interesting one to come another change how do we develop code we've all seen and even in this conference you know a lot of talk now about containers and we've seen actually you know the whole VC world and everything else get rally around a lot of new companies who are looking at the code development process from the time you sort of have an idea and and how you can package that up can you share components with other other parts of the industry you know in the in the electronics industry itself those of you who ever did like circuit board work or VLSI design or whatever we had standard libraries and we had components and reusable pieces of hardware and we're trying to get that same kind of reuse now and something which is kind of containers but we've seen a lot of things now start to come out here so it's not just been docket but Kubernetes and mesos and core OS so we're starting to see a different layers of this come together so it has layers and we saw even the demos during the during the keynotes of now application libraries where you can bring these things into your system very readily you can containerize these and you can deploy them and at the same time we're seeing operating systems are getting skinnier and skinnier because they don't have to support now everything in the world they can really support much more narrowly cast applications so I think this is a very exciting area that we see and a lot of people when they're looking at this look at it so you know it starts to look like a cloud so in fact a lot of things if you look at mesos and another platform such as that it's a distributed operating system so it's an operating system that is running actually on top of an operating system but it's distributed across a data center and it has resource scheduling capabilities in it you can say I want to have five copies of this application running at all time and it will use Zookeeper and other things to make sure that that application only says five instances of it running so if one should go away it can automatically start up another so a lot of the distributed computing work that's been done over the years in different systems I think it's turning the sediment down into these new layers making it again easier and easier for application developers to get their job done and then so as we start to think about this the question is is there a battle here many people right now and I think that you see even reports out there is that you know doctor and containers and mesos and Kubernetes is going to kill off cloud computing and I actually don't find that ever to be the case with technology technology has a wonderful way of taking the best from every different piece and use it for the most appropriate use so I think in some up some uses absolutely we fully just containerized but then it's not really designed for the whole kind of multi-tenant notion that we have in a cloud so in fact if you look at the differences between virtual machines and containers there's some very primary ones and and so virtual machines we think are fat they're heavy so on top of a host you're running now an operating system and depending upon type one or type two hypervisor that allows you to create a virtualization layer to live vert you can talk to that and now you the customer is bring a virtual machine on top that virtual machine thinks it's running on hardware because what did we do with the hypervisor we're emulating a hardware layer so through that emulation it makes it possible to run an operating system and an application let's say which might be Linux next to another one that's running Windows because we're emulating at the hardware layer and this is this is the brilliance I think of virtualization and allowed us to do the stacking of application and dramatically increase utilization of of our servers containers on the other hand really do this at the operating system layer itself so here you are able to do segmentation of through namespaces and C groups and others so that you actually are running simultaneously two different applications that sharing the operating system by and large but may have different user libraries may have different libraries in terms of which level which python release you're running that allows much lighter weight sharing of resources greater utilization in terms of the memory overhead you're not carrying around the whole operating system with you wherever you go but again it doesn't now allow you run Windows and Linux applications on top of a common operating system because there isn't one so they each have their use it's important for us when we're designing applications to know when we want to use one technology versus versus the other what's even more interesting is that when they're used together and what I'm particularly intrigued by so you've seen work through things such as Magnum where we're talking about well what if every tenant what if we can have a service and a tenant comes into the service says spin me up a Kubernetes environment cluster Kubernetes cluster our dark Docker swarm cluster and be able to spin that up and that's owned by that tenant whereas another tenant can bring in their own cluster and again running on top of virtual machines so this is where the cloud computing model and the the container model comes together in fact when you're running containers if you've done an experiment you know an Amazon or Google that's actually the model you're using we can talk about then those virtual machines as getting smaller and smaller and smaller but they're still providing there's still hyperbop hypervisor based isolation between virtual machines but they can get to be very small what's interesting with OpenStack particularly in the Kila release now we're also really pushing and bare metal hosts and so through Ironic now with some things like Magnum you can start talking about running these on top of bare metal which essentially means all you've done is sort of partition again your host resources to buy tenant to giving them capabilities of running a pure kind of container environment so Magnum is one of the projects I urge you to take a look at and get involved in we're heavily involved in this from Cisco's point of view and this is really containers as a service according to the model we just talked about where there are APIs that allow you to spin up these container clusters and that it supports both Docker swarm and Kubernetes and we're using heat again to provide the orchestration of these services and they have as with I think Marcoli also mentioned you know this is running on what's known as a bay which is where then the environment for the containers operating system works and this allows us to basically use both kinds of of image-based deployments at the same time another interesting services Kola this is where we're looking at what if it one of the issues that we have I think in OpenStack as we all know is the manageability and availability of these services and so what if we use some of the technology that's coming out of container world to help with the manageability of OpenStack services itself so in this way a lot of the work that we're doing here is to run the OpenStack services Nova glance and they're inside of a container that allows us to put the container in a cluster of machines and allows us to do things such as more easily run one version of Nova upgrade from Havana the kilo at the same time and then start to switch things over so in this kind of a model I think we're looking very deeply into this to understand is this a better model for how we actually deploy the OpenStack services themselves and so again I urge you to take a look at that project all of these oh and this is just a mile one of the interesting thing about deployment of opensack if you think of the application space and what we're doing with CI CD to deploy applications we would like to do that for OpenStack itself so the CI CD pipeline becomes very important and that again feeds very well into the whole model of containers containers are meant to ease the whole integration deployment rapid deployment of applications that application changes in a very very lightweight fashion and this this allows us to do that for OpenStack services itself that's where a lot of the power is going to be so with all of this we've got a larger number of things that we are doing and we've got a larger and larger systems in different kinds of application spaces so this is where it becomes much more imperative to really talk about how we're getting the stats out of the system being able to use analytics being able to dynamically create these things and use visualization so some of the work that's been coming out of some of my team also you can check is AVOS where it really integrates both visualization and advanced analytics capabilities into running on top of OpenStack so you can get a better view of what's going on and being able to run some of these services both at the system level at the application level I think is important as we increase these number of services and microservices again we have more and more need for this so one of the simple ones we've been working on lately is something called cloud pulse and we just start launch this the other day as another project think of this as a monitor a health monitor for your OpenStack cloud so that just like an intensive care you go into any kind of hospital wherever you always hearing these like beep beep beep that means the patient's still alive the cloud's still up if that ever stops or goes to a high-pitched wine you know your cloud is in deep trouble and so cloud pulse is where we're gathering a lot of these metrics and again working with things like Monastka and everything else being bringing those together so that you can have a health check of the system because and provide this health check again at a much more detailed level for the application providers because the worst thing you want to do is running an application on top of a cloud and you have a problem and you go is it my problem or is it a problem in the cloud so by increasing our level of visibility around the analytics that's taking place in the cloud platform itself will make it easier for application providers to understand whether their application is at fault or whether it's something going on in the cloud and as we get closer to that understanding what that fault might be so that they can take action and it shouldn't be to call up the cloud provider or your IT department and say I've got a problem here you can increasingly use automation then to get possibly move virtual machines out of an area that's having a problem and move them to a healthier place so opens that keeps expanding and this is viewed sometimes as a you know benefit or a curse more and more services are coming along I think that we've you've probably heard talk about you know this big tent concept where we want to have more and more services become available provided they are adhering to the principles that we have around open source and so this might be confusing at some place but I view this again as opportunity and that what we have is that new things that are coming on in terms of data processing and morano is an application catalog that we're now seeing being tied in in what we're doing with containers this becomes a much more more powerful platform that we are working on so the real challenge here I think is that with these ever growing number of possibilities and opportunities is essentially can we keep it together I mean can we find the right balance between simultaneously moving towards carrier grade enterprise ready solid resilient cloud platforms as we continue to expand and grow the number of services I think to do that we have to take much more of an overall architectural view of how these services put put together and and really you know focus on what is going to ultimately end up with the true value to the end user customer by the way I'm just getting the sign to close off but I think there will be time for questions so if you do have a question or whatever we do have a mic up in here before I close so more than anything what I'd like to do as a community is to find that balance and work on both of those layers both towards increasing resiliency and availability of the platform and scale at the same time as we don't to accomplish that we shouldn't shut off all innovation because I think the innovation that is coming is very very real and providing very very high value so finding the right balance between those two and those of you who are contributors into into the open-stack community it'd be great if you really recognize that you should be working on both those things that way also as we bring in new services we want to we want to emphasize those services that are actually increasing resiliency and availability of the platform that's the ideal world so with that I'll try to open it up for questions but thank you very much questions anyone yeah or shout it out I'll repeat it will get okay so let me I'll repeat the question so Cisco is known for its routers and switches and proprietary systems essentially even though Linux has been used inside of Cisco's routers for years and years so we've been a big consumer of open-source software I think that what we're seeing now is that we want to be much more of a participant in the open-source communities and your question I think is around so how does that translate into a business model for Cisco because of Cisco is helping essentially putting all of this technology out back into the open-source anybody can use that so how does that create particular value for Cisco in a business model I think many companies are struggling with that same question and I think that we are all recognizing though there's something new going on here in in kind of community-driven development standards bodies have existed for years and years because the industry recognized rather than inventing 25 different ways to plug into an electrical outlet we can make that more money in our individual companies if we standardize on those places where you interconnect so interop has only been driven by standards well the new interop now is when we're creating software platforms is for the develop the applications that run on top so I think Cisco very much believed we are working with customers many of our customers who are deploying open-stack they are buying a lot of Cisco gear to do that and we are providing our expertise and support in there we also have provided a lot of plugins which allow us to integrate into Cisco systems so we fundamentally will be competing on open-stack works best on Cisco Cisco is not just about routers and switches we also have servers and storage so we very much want to provide assistance and and capabilities back into the platform that will make open stack work great on top of an integrated environment delivered by Cisco we also think it's just very important to to have this kind of community-driven development I think that if you look at the number of companies below the BHP Intel Dell that we are all working together and we've decided it's better for us to work together and grow cloud computing than it is for us to create a million different kind of cloud standards out there which I think will hurt the industry long-term follow-on questions unless somebody else is willing to stand up and here we are so Cisco's been at 65 percent margins I think I know where you're going with this yep okay well as a Cisco investor let me address that so yeah we're 60 with the with the business model that sort of demands that we maintain very very high margins you have to create very very high value that's the only way you can do that so for example what we're doing with our inter cloud strategy is rather than Cisco standing up our own clouds and making all that we are making that investment with service providers around the world because what we see Cisco has a global reach guess what we're a networking company we connect the globe so it makes perfect sense for Cisco to have a strategy around cloud computing of creating a global network of inter cloud nodes that's what we're essentially called so we are making a number of announcements that we've made about partnerships with many of the largest telcos service providers around the world where Cisco is going to be building a cloud they're connecting these clouds together and they're primarily their focus there is however not to compete directly with Amazon and even Google which can do this a huge huge volume and and instead our business model is to focus on the business applications so it's much more about providing the business services on top of those those services are largely for many of those telcos they are business networking services their connectivity its VPNs if those kinds you know direct data links it's those kinds of services and then our own businesses around collaboration and around security many of Cisco's new businesses are cloud-based businesses you look at morocchi how we're with morocchi that is being that is managing Wi-Fi access points from the cloud so we see our future growth is in terms of these cloud businesses and that's how I think they will be able to maintain a successful company times out I will be available out in the hall if you want thank you very much