 Good morning everyone. As Mark was suggesting I work at AT&T. I suppose you were probably expecting Locutus of Borg or Grand Moff Tarkin or the like accompanied by the Imperial March. At least some sort of suit with the big big old telco kind of mentality. Dubious intent and hopefully hopefully I'll disappoint you that way. Anyways so AT&T when I first arrived as a part of an acquisition I was very surprised to find that open source was pretty much forbidden. We're not allowed to use it. It was only used in very small places and I applaud the efforts of those folks that continued to you know carry the flame of open source. Folks like Steven North on GraphViz and the people that worked on R during this period of time they kept that flame alive. It was surprising to me given the history of AT&T and our involvement in Unix and see kind of the the integral pieces of everything that we do and especially Dennis Ritchie's contribution to sharing giving the tapes of Unix away to other companies. This was a seminal moment in the beginning of forming formation of open source. I've always felt as a part of AT&T that I could help make this change happen given that I've been a part of at least an user an abuser of open source for a very long time. Felt like I could bring that notion into AT&T. AT&T now has changed its thinking. Today it's much different place when it comes to thinking about open source and it is now an integral part of our strategy. We are active participants not only in OpenStack but in other projects and we're now looking at expanding that into a new area that I'll talk about today. The NFV open platform is in example of this change in strategy that is the result of open source and specifically the result of our work on OpenStack and our contribution there. It's taught us how to share again. If you look at how we how we typically work our vendor relationships are very much about very small number of vendor relationships because it is very onerous to create a relationship with AT&T. Many of those relationships have lasted over a hundred years and those are very large and it takes quite a bit of effort legal wise and negotiation wise to get those in place. When you look at the 400 companies involved in OpenStack it would take us to 2112 to be able to make agreements with that many companies. OpenStack has provided a proxy for us to get innovation into AT&T that's very unique. This is a concept that I feel very strongly about. It creates an ecosystem that we can work with and get help from from a lot of different groups. Our contributions have not always been that great in this space. It's growing but indirectly with our partners we have made a lot of contributions and you'll see that with what we're talking about today. OpenStack we've been using it we've had it in our lab since 2010 I was at the very first summit. It's remarkable the growth that's happened here. I remember being in a one-room summit with not very many people and then now it's honestly very daunting to be talking to thousands of people in one room. The change for us as we started using it in 2010 in production we actually put production workloads into end of 2011 we today run a number of websites like itcanwait.com something I highly recommend you go to pledge AT&T's developer programs around its API exposure our mHealth program our toggle our dual persona work the back ends for our address book that's on everyone's phone these things are just the first steps that we've taken into deploying applications onto OpenStack today it's about 120 applications now we've lived through good times and bad times with with OpenStack obviously for us we actually still have pre Diablo versions cactus versions of in production which is somewhat problematic and an issue that I know the community is working on I'm very excited about a nice house the work done to get to rolling upgrades and the promotion of CI CD and these types of things in OpenStack this is is very helpful to us now as we've gone forward we've expanded our footprint so it started with three three data centers and now we've we've deployed it into seven data centers today at this end of the year we should have we should add three more it's the use cases that we're using have expanded from typical back end applications and some of our external services are API gateways they've expanded now to include new things especially big data and and and then a whole new area of function that that I'm very excited about it keeps me coming to work if really pumped about the opportunity over the next period of time over the next two years we actually plan to expand the footprint from seven to 10 sites and expand that by 10 to 20 sites yes you might ask why so many locations the answer is NFV or network function virtualization so network function virtualization hate to be a purveyor of yet another acronym AT&T is an acronym and it's really excited about its acronyms and feel free to throw something at me if I reel off 12 in one sentence it's very possible when we talk about the the mobility systems that we work on but NFV is is making a significant change in our industry it's causing galaxies to collide the traditional IT space and the traditional telco space of function and competitors and vendors they're colliding they're colliding on top of a concept that is essentially the ongoing onslaught of Moore's law of automation of agile methods of consolidation commoditization these web scale these concepts are now catching the systems that we run inside of our our network the things that we use to run our mobility systems and the things that we use to run our TV platforms they're all getting caught up in the same desire desire to be more agile to move faster to to be to have higher utilization to lower the cost to move faster to be have bring function to market quicker it's those things are getting wrapped up in it we even AT&T is having competition and from unique new part new players and we have to move faster we have to change open stack open source is helping us to do that because I believe that open stack can expand to include workloads like NFV so again this this concept of NFV is really changing the landscape today in our networks we run a number of things that are built on very specific hardware things that you may be familiar with routers and switches firewalls load balancing all things that have been traditionally things that have been already running on open stack today already part of the SDN phenomena and then beyond that we're talking about elements of the mobility system things like session border controllers or or huge NAT devices they're they really are custom built very expensive pieces of hardware but when you look at them they're actually very similar to the other things that we run they look very similar to load balancers and firewalls and and application servers they and thankfully they come to us with natural pre pre dereliction toward scale out toward cattle based architectures so in a way they're ready they're ready to make this transformation and if you look at this space you see a lot of open source players that are coming to to to actually solve this problem starting with the things like asterix and such starting at that point those are expanding to include many of these network functions both on the mobility side and on the TV side as we look at and then are inspired by what Netflix has been able to do on a public cloud in a very scale out model we are able to be inspired to change the way that we we deliver TV content to our to our uverse customers so this this is helping us in a number of ways obviously it is by changing to a new model taking everything that was running in specialized hardware and moving it into a virtualized world either VMs or containers or even virtual appliances bringing them into a multi-tenant space allows us to get higher utilization more agility and allows us to be more competitive and allows us to distribute function closer to the user the real advantage that AT&T has them if you think about it we have an enormous number of physical places around North America where you could put servers and by doing so and and having a lot of underutilized bandwidth between all these locations if you think about it we can take like content distribution to the next level make it far more dynamic and have it exist closer to the user this is essential to things like moving a mobility system into our into a virtual platform taking and making OpenStack handle more distributed methods is a key part of making this transition happen so to me OpenStack is well positioned to handle this this new workload the things about OpenStack that that I like beyond what I was describing as a proxy for working with a much richer ecosystem a plugability model this plugability model in like sender and in and a neutron helps us to get leverage over vendors and to work with a broader set of vendors and within that model there is enormous peer pressure between vendors to have high quality code testable code stuff that actually works and is scalable and perform it beyond that OpenStack and its innovation cycle it's helping us to promote the use of agile methods within AT&T it's an example of how continuous integration continuous deployment and these types of things especially our work around tempest and around ref stack and def core and these things they're helping us to promote agile within AT&T a place that pretty much invented the waterfall method now this innovation is causing us to really question not just within this space but in other spaces as well SDN is a particular example where we're now realizing that the routers and switches that we use they they need to be virtualized as well unfortunately the SDN term is very overloaded and when I first heard NFV I thought do we really need to have another term since SDN is already so overloaded with hardware software disaggregation control plane data plane separation and hypervisor switch vSwitch virtualization and overlay protocols and orchestration and automation why do we need to have why not just add in the the layer four plus functions and the things that telcos think of as network really applications why not just call that all SDN unfortunately my co-workers are the ones that came up with NFV concept so eventually I have to have to kind of comply with this this idea and in some sense it makes sense because NFV means really a different kind of application footprint a different kind of function so hopefully and you know when I look at all of the benefits of open stack and where where it's been and where it's gone and and the next nine releases as Mark was talking about I'm very excited about it the opportunity it's it is awesome now the thing is and this reminds me of when I was a kid I had this bike it was called the Raleigh chopper mark to was it had a small wheel in the front a large wheel in the back very large tires it had a banana seat with all leather for the nice back had a sissy bar and it had shock absorbers it had a like a muscle car shifter and with three speeds had eight hangers it was awesome it was red unfortunately it suffered from two fatal flaws one that the shifter was not really in a very strategic location and when you were going more more onerous was when you were going downhill at high speed and you applied the brake pressure equally you had a tendency to reverse wheelie and fly out the front of the bike in this process I lost a tooth a cleft chin part of my face and mysteriously at the after the second time the bike itself disappeared don't let open stack have have this kind of fatal flaw people say you know the open stack is fatal flaws that it doesn't have a benevolent dictator people who say that don't understand the model the ptls the projects they are essentially separate open source projects that have a unique federation a unique binding cohesion that is the overarching concept of open stack that makes it work that isn't it's fatal flaw this fatal flaw is not scaling from the level we're at now to the next level I am confident in the model I'm confident in the people and I'm confident in the adoption that's happened for me the the fatal flaw is at a lack of you know expanding the paradigm thinking about open stack more broadly you know not every workload is is a perfect scale out cattle application every once in a while one of those cows look like Bessie and you kind of fall in love with it and unfortunately that can give you a black eye the pets have to be handled and have to be managed the enterprise workloads I feel have to be brought into scope more more clearly and then on the other side when we talk about catering to telco workloads reliability performance scalability distribution to many locations has to be a part of the story it has to be included into what we're thinking about and in the end I think open stack can handle that and it can expand to the scope so hopefully if you'd like to hear more detail more specific technical detail about what the changes and specifics about NFV and SDN and then some of the work that we're doing blueprint-wise and changes we're trying to get promoted into open stack to make make it more usable for this new NFV workload please come in and hear Mott Carlson at Ericsson and I talk about about this at 250 thank you