 Hello. Hello. Hello. Good afternoon, everybody. Thank you for attending our session this afternoon. It's going to be a very exciting session. I'm speaking slowly so that it can be translated more appropriately. But what we're going to do this afternoon together is try to predict the future. It's often hard to predict the future. However, in this case, it might not be so hard because we have some colleagues of mine from IBM, but also leaders in the open source and in particular open stack communities. And as you know, oh, hi, Monty. How are you doing? Good. Very good. So we have leaders in the open source and particular open stack community with us. And so the reason that I don't think it would be hard to predict the future is that these folks here are actually defining the future. They're defining the future of where open stack is going to go in the next six months, 12 months. But also, I think more importantly, working with a lot of you all, our clients, our partners on implementations. So for the next time we have, we will be doing that. So what I would like to do first is ask each of you to introduce yourself and then we will continue with the panel session. Sorry, I got to steal your mic here. So hi, I'm Monty Taylor. You may have remember me from such things as running around on stage in a awesome sauce costume this morning. But but also from being sort of in the open stack community and ecosystem since since the conversations where we were deciding what to call it. So I've been around, probably have met many of you in other times and thrilled to be with IBM. So I'm Jesse Proudman, founder and CTO of Blue Box. Blue Box has been working with OpenStack for I think three plus years. Blue Box was acquired by IBM in June. And as you saw in the keynote today, we launched our Blue Box local on premises offering yesterday. And I'm Mo Abdullah, and I'm actually part of the product management and strategy team that really works with OpenStack and the rest of the open cloud architecture building products that take advantage of those technologies. In some ways, my perspective is one of gratitude to the broader community because a lot of the products that we bring to the market leverage the collective. And then the other of course is the give back aspect, which is how we take a lot of what we learn from a commercial side and continue to grow and nurture what we're doing out there. My journey with OpenStack started from early days when we really understood the power of unifying the disparate, if you like, architectures and thoughts out there into one that brings multi clouds together. Glad to see it grow and reach the point it has so far. Hi, I'm Mark Jones, CTO at software focused on infrastructure. So we've been running OpenStack Swift for about three and a half years now in software data centers. And we spent the last nine to 12 months working on bringing OpenStack natively into software data centers. And so we actually launched our VM service yesterday in the London data center via Blue Mix. Yes, I know. Step one, it's a great effort across many teams within IBM to deliver that service. And that's the first of many announcements we'll have this year and well until next year. Okay, so I'm going to start with a very simple question. And I would like to ask each of you to please answer it within a two to three minute time window. I'm not looking at you. Is there a competition by the way? Yes, there is a competition. It's always a competition. I'm going to start from that side in. And the question is, can you give the audience one prediction as to where you see OpenStack headed in 2016? It could be a dimension around a particular project. It could be a technical dimension. It could be a client dimension, whatever you like. Let's start with that. So I'll start with the project that is near and dear to my heart at software, which is bare metal. And so that would be the Ironic project. I think in 2015, we've seen Ironic mature. We're seeing implementations of Ironic make it into production in everything from companies that are deploying application stacks onto it wanting the performance that you can get out of Ironic. And we're starting to see it now in service providers. And when we see the evolution of Ironic software, one of our core differentiators in the marketplace is bare metal. We've done that since inception back in 2005. And we are putting a tremendous effort into Ironic this year and into next year. I think when you see the evolution of compute now with containers, a lot of companies are looking to have near bare metal performance. With the ease of use and management of containers. And so Ironic is going to be a project that's going to evolve quite a bit in 2016. Thank you. Mo. My prediction is that next year we probably will see an incredible amount of growth in clients who are going to be talking about their stories, leveraging open stack in production. We'll go from simple hundreds to many hundreds who will freely share in Austin and conferences after that. And some of the drivers behind it, I believe is simply the maturity that the community has taken the technology forward from simple discrete components to things that can integrate together through services like heat. So as my team, for example, and many of the leaders in the room sitting in here, start to build public cloud services. They're starting to leverage things like Neutron and Swift and others and put them together in patterns that serve the end consumers in applications. People will start to talk less about the individual services and will start to talk more about how they're putting them together to deliver specific client value. And I think people like IBM and others will come forward and start to talk about our journey and in the way we're standing up, highly scalable, globally distributed services that are in their core and in their heart built on open stack. So thousands of stories and much more oriented around combinations of services. Thank you. If we were on the prices right, I would say there's thousands plus one stories and I would win the car. My prediction is that we'll begin to see different compute primitives exist and be supported and open stack more natively. So Nova has supported VMs and done that very well for a number of years. Certainly with ironic now as Mark mentioned, we've got bare metal and you see the transition from Nova bare metal now to ironic. And we're seeing the same thing now with containers with Magnum. So you had Nova Docker originally now Magnum. And I think there's something very compelling and powerful being able to abstract bare metal VMs and containers all within one set of APIs, one authentication domain, one network with the work going on in Neutron. And if you look at many of the application patterns that are being deployed, it's that combination of services bare metal for IO databases, et cetera, containers for certain aspects and VMs that make up modern applications. And I think that's where that's 2016 will be the year of support across the board from a production perspective of all of those capabilities. So I couldn't agree more with all of those things, which is why we're saying also you're all right and I'm going to add something else on it. So that's my 1000 plus one story there Jesse. But I think the key word that already sort of came up that I'd like to focus on is the global service part of this. I think in 2016 we're seeing the things we're doing with SoftLayer VM and with Blue Box and SoftLayer, the key part of that story is the global footprint of that. So as we look at public cloud service and we look at managed private cloud services for people, it's not just about having a data center in Dallas or San Jose, it's about having a data center in Singapore and in Sydney and all over the world. All of us, there's more people in the world than not in the US than there are in the US. And so the highly US centric nature of the offerings that have been out there so far, I think we're going to start seeing the real power of a sort of global first approach to public and managed private clouds. I'm going to be really excited about that. This is in this short amount of time you all have covered four amazing topics and I think we should write this down and check it next year. But think about it, the power of bare metal applied to a variety of programming models because of the utilization that you can get, the performance that you can get. We've proven that out in our SoftLayer model already and now it's going to be to the masses with the work that we're doing around it. Mo brings an amazing point around moving from atomic elements to composites of elements. I mean perhaps even our OpenStack marketplace might evolve to a pattern marketplace where there are literally hundreds and thousands of pre-built compositions that can quickly be provisioned, managed across compute, storage and network. And then of course the confluence. This morning at the keynote one of the points that Jesse and I made and Jesse made in a much more detail here is literally in cloud there are three centers of gravity right now. There is OpenStack, there is the cloud native computing foundation containers and then there is Cloud Foundry. And if you think about it, OpenStack is the basis, compute, storage and network. And then there are two fundamentally complementary application programming models that are changing the way folks develop applications. One, Cloud Foundry, a polyglot approach to building applications quickly, systems of engagement for example applications and then microservice based on containers. All using the same substrate. I can certainly tell you that these folks here and our colleagues in the community are going to be working really, really hard to ensure that all of these things come together. We will not duplicate in say the cloud native computing foundation network services we're going to use, Neutron for example. Same thing in Cloud Foundry and everything else. So I think that's a pretty state. And then of course globalization is a necessity. You know the era of the mega data center really is dead. It is about having data centers fit for purpose where you need it. Dedicated across the globe. We certainly are clients. You all have demanded that. We've been doing that. I think others are starting now to try to catch up there. But it is a huge deal. Those are four very, very solid predictions. Now what I'd like to do is offer up some perhaps more controversial or less controversial predictions and ask you all to debate them. I got some questions from the audience prior to this. No I didn't. I did not. I do have to say though I wish Sean wasn't sleeping over there. Thank you. He's distracting me. First question and what we've heard this theme several times and that is the distribution is dead. Okay. Discuss amongst yourselves. Jesse you have something to say about this? That's my tagline so I'm not sure I have a whole lot to say but actually I do because that's what I do. So look open stack is a challenging set of technology. It's challenging to operate. It's ever growing and expanding the velocity of new projects and of the complexity of those projects. It doesn't stop and it won't stop if we are going to continue to maintain relevancy in this marketplace. And if you look at software and you look at software distributions and Linux distribution the blast radius the amount of failure that you can tolerate when one server goes down it was much smaller. A single system may have contained a database or a application server and that was fine because if you designed your applications in a certain way you could tolerate that. When your cloud goes away for many organizations that's their entire business and so selling software that pushes that operational burden onto a customer is unreasonable. If you look at the growth and open stack from a revenue perspective and where most of the installations are actually occurring it's from a service provider perspective. Because we're centralizing the expertise that it takes to track these projects install these projects operate these projects and upgrade these projects into one central provider. Further you get this concept of a network effect where for every failure that does occur and they do happen. The learning from that failure the fix from that failure gets pushed out to every other customer and can be pushed out in real time in a managed service model. And that's something that just doesn't happen with a software distribution and so from from my perspective and I am mildly biased. I believe that the services model delivering open stack or private cloud as a service whether it's in software in a dedicated approach or in a customer's data center with our local offering will make customers more successful and make that success occur faster than with a software distribution. And I've I've not found anybody here who can argue that with me. So if you disagree come argue with me. I'd like to agree and sort of poke in a mildly different direction to we look at things like like Docker and the work that's that's going on there and and and you look at what the developers are doing with their applications. And most developers these days like even outside of cloud they're not focusing on delivering Debs or RPMs of their of their software they're delivering Docker containers delivering that application that that they made in a in a consumable form. The the the sort of distribution model we've had for the last you know 15 some odd years has worked well for us. It's been a it's been a it's gotten us where we are you know. But it's predicated on the idea that you're going to have a number of pre a number of building blocks in the distribution. You can sort of mix and match and install any of them you want to. So you can say after get install XML server and and then also a postgres database and and a whatever and those should all kind of work together and you can expect that one of them installing one isn't going to break another one and that's that's great. But to Jesse's point the software really caring about these days is way more complicated than those tools really really give us the the the benefit to to get so I think even not only in cloud is Jesse right. But even just as we're looking at how people are developing software targeting that software being inside of a distribution isn't it's not how people are thinking it's not how people are thinking about solving the problem because it doesn't it's it's more overhead than than it is for the help. So I think that the reality is we've seen this transformation between traditional software that is purchased to software as a service right that transformation occurred delivering cloud services or open stack as a service is no different. It's the same models it's centralizing that operational responsibility into a single party. And I think the conversation changes frankly not that it's not a point of debate it's a point of agreement around SLAs and really it's not about the supported model in which I'm going to get a distribution but it's more around what SLA am I going to get how can I work between providers it's not dead fully yet as we agree because it's a journey but it's dying it does not lend itself well to the go forward model. Okay well you know it's you know by saying the distribution is dead doesn't mean that private cloud is dead what it means is that you want public cloud capability if you want on the public cloud on a dedicated or on premise locally right and you've coined the term you know private cloud as a service right whether it's on premise or dedicated you know the blue box local that you discussed supports that model I encourage you all to try it it allows you to more quickly get time to value. We have some relay technology which makes it very easy to do that and I do fundamentally believe that this is another layer of abstraction. We think about computer science for a moment if anyone here is a programmer you know there has been a constant democratization of technology we don't program an assembly language anymore right we don't strictly right define you know we just we don't use methods I mean sorry functions all the time we use classes and objects and we expose data right for those of us more sophisticated we might even do inheritance or multiple inheritance right service oriented architecture is object oriented network we don't debate that right there is a democratization of technology and it's happening again now and I think the quicker we do that means the quicker we can add value right and that's what we're here for we're about innovating on things. So another question I have here that that I'd like you all to debate around the future is around interoperability one thing that I'm very proud of and I know you all too is Catherine is our PTL for RefStack for example she's been doing a great job and this is about interoperability and how do you see this playing out next year both in terms of what we actually test okay because right now it's not as much as it should be and in terms of people stepping up to actually adhere to these things. So we've got a we've got a saying in in OpenStack which I think originally came from Russell Bryant but which is that if it's not tested it's broken right and and one of the things and one of the reasons that the RefStack effort has been so important is that it gives us an opportunity to to be able to test and vet what people are doing with their clouds and also a way to express what behavior we expect them to do because we did create a flexible, pluggable tool for for people to use and and so it's it's not unfair for people to have made some different choices with it we we kind of we kind of built it that way it's pluggable but but that same time that that now needs to come with a thicker instruction manual a so yeah you can do anything you want but actually here's some things that are really important that you that you keep in right. Right these are these are things that our users are going to demand to exist and some of this is things have changed since since five years ago when we started off there's a session about glance image uploads tomorrow in the design summit and one of the things in the in the in the write up for it the guy who's is leading the session talks about when we when we first envisioned glance no one envisioned it as an end user service they envisioned it as a service that Nova would talk to to manage images and now we've all realized no that's actually really important cloud without image uploads is useless right it's not a cloud that's that's new information and all we have to encode that ourselves and also make sure that we're telling everybody that's running a cloud no this is really important seriously and most of them once we're really clear once we make a really clear statement about this is an expectation almost everybody is has been really interested in complying and in making sure that they're everybody wants their product to be something that users can or not operate with like that's that's what we're all in this game for and so I think that the work we've been leading there and the work that that is continuing to move in that direction is is absolutely essential to empowering all of the people in in our ecosystem to play by a set of rules that that's that's beneficial for all of our our users and customers. So I agree with Monty and I think from a product perspective if you look at it in a mark reference the the object store beta. If you look at what clients are trying to do is you can never box any of them into a single pattern. Different people are starting from different points and one of the things that are unique around object store and the ref stacks and the compatibility is also at the API level is they may start with building applications that take advantage of an object store and they're testing those in a market and then they decide that they want to geographically expand. And this is a true case in several of our clients who decided to expand into Europe. Now just a simple you know privacy rulings etc dictate that their application now needs to talk to you know local object stores that whole idea of being able to point your application away from a public object store right into one that may be running inside of your own data center managed by Blue Box then you know without you know changing the application logic etc and with speed is very powerful very compelling. I recall actually the concept of this we did about six months ago in Vancouver and I recall everybody laughing at me and going more this is like so simple. Like this is even more simple than somebody shooting crocodiles you know on the screen and I was just like but look it is so simple but powerful and the room was full with people who are saying that's exactly the type of scenario we want to enable. And then you extend it further into different application styles. So it's you're building these applications based on containers virtual machines. How do you actually recompose those in different environments. You know I think one of the things that that's challenging. It's both a pro and a con of open stack is that it really is a Swiss Army knife. There are so many projects and with this big tent initiative that that number will continue to increase. And that becomes a challenge to the cloud operator whether that's Blue Box whether that somebody has bought a distribution or somebody that like Walmart or eBay has decided to do it themselves. You know I think that the key notes on Tuesday from Bitnami where they were saying it's very difficult to figure out that consistency of services from cloud to cloud. That's going to be a problem we continue to see because of the flexibility of open stack because each deployer can choose the projects that suits what they're trying to do. And while I think a huge fan of ref stack the reality is having consistency in services across every cloud it's a challenging proposition. And so that's where I think beginning to see providers in the marketplace that have all three consumption methods that have a public cloud or dedicated cloud a local cloud public and private and having all of that speak the same technology. You can get that from one vendor and there are a few vendors in the market that can do that as of January 31st. There's one less but there were a few that's really important and it will continue to be more important as we add more projects into open stack. Thank you. You know I encourage you all to demand the expansion on ref stack to demand the expansion on what we define as interoperability. Todd why are you smiling. Todd's on the board for us here you know so he's been carrying that flag for us. So he's doing his part but I demand all of you to demand the demand I ask all of you. I'm a little tired I apologize. I ask all of you to demand that the expansion of ref stack and also to demand to your vendors whether it's IBM or anybody else that they adhere to that because interoperability is going to allow to succeed. You know our job is to make all clouds behave as one not just IBM's but all of them and that's how you get the time to value you need. This next question is for Mark and for Monty or anybody else but I like to start if it's okay with you too. So let's put a silencer on you in the middle. And the question really is you know it amazes me when I wake up and I see open stack available and what we're doing at a global scale. Okay 40 plus data centers across the world is unbelievable. Not just global scale but scale scale. And you know as we do this we find places in open stack that need to improve that need to change in order to actually manage the scale that you and you are driving together for our clients. Can you give a prediction on some of the areas where you think we might have to change a little bit some of the architecture based on our experiences so far in doing this. Yeah so I think in the work that we've done so far that we see an area for improvement is going to be Neutron. We start talking about scale and performance of the network it is the key linchpin of everything that we do in cloud right. Everything rides on the network and that's everything from initial provisioning to configuration to throughput and resiliency. And so I think Neutron is at least from where we are today is the one that we see a lot going into. But I think you can apply it to every project. We start talking about as Angel mentioned we've got 40 data centers. We're initially targeting three data centers for software deployments or open stack but we will extend it to all data centers. And that becomes global scale. And within each of those data centers you have to have full scale up and scale on capabilities. And that's a large footprint and customers demand resiliency. They demand uptime and performance. And so I think that every project is going to we're going to find stuff with every project as we go. But for my standpoint I think Neutron is going to require a lot of work. So a couple things. One I actually want to point out a thing that also happened in yesterday's cross project design summit session. One of the design summit sessions is there was a session on starting to map out supporting the extreme use cases as they talked about it. We were really good at 100 node open stack clouds these days. But doing you know say a million node open stack cloud not maybe not so much. And then also for the sake of that that conversation was also a question about should we also be focusing on one node. You know supporting people who want to do one one node things. Nobody in the room actually various their hands and said they're really interested in the hyper small scale. But it's worth asking the question to make sure we've got use cases. But everybody basically signed up to agree that supporting actually supporting the the hyper large size things is is an endeavor that the technical community does think is is important and agrees that it is a thing that has to get attacked. So that said things that I think that are inside and I apologize to anybody for I'm going to geek out on YouTube for just a second. I because I think that the things that that we're going to have to attack are actually seemingly very boring non sexy internals bits of of all of the services to Mark's point. It's not just one or the other because we've only been focusing on on. We haven't been focusing on 100,000 node A Z's than the challenges of resource management and consistent cluster management of the resources that open stack. Manages has been left as an exercise to the to the deploy right. It's you would give you a bucket of parts and you you can do the thing that you need to do and that that's fine to a certain point after a certain point. We've got to have more primitives inside of open stack to allow the deploy or to even manage to keep up with it. So we had a we had a great session yesterday on distributed lock management which is probably going to put half the room to sleep here. But a great great session on distributed lock management and using technologies like zookeeper as a fundamental building block for doing resource management and leader election. These are all things that are that are only partially looked at inside of open stack right now. And so this is we've got a couple use cases but this is going to this ball is going to start rolling faster and faster forward. So getting those things turns out scale and manageability kind of kind of go hand in hand if we just throw 100,000 machines out there and the operators can't keep up with what's going on in them. Then it's you know, it's a it's a it's a it's a lost cause. So these are these are areas where we're really going to stick a lot of energy and make sure that we're we're enabling us to get to those to those next generations, you know, orders of magnitude of scale. And I think that, you know, a lot of the talk is around workloads and the type of deployments that we're going to put on to open stack. But when we look at these these large scale deployments, there's a huge difference between running a public cloud, multi tenant service built on open stack versus running an application that you're able to fine tune and architect to meet the needs of application. And so as we as we go global scale, as we go these incredibly large deployments that are multi tenant, we learn new things that you don't necessarily find as part of an application deployment. This is critically important. And I, you know, I ask you all to take note of that because there will be some exciting changes as we really transition open stack from thousands of nodes to millions of nodes. Okay, so I'd like to end the panel with one question for each of you again. And I'd also like to ask the audience to take your phones out and use your favorite social media device, whether it's Twitter, Instagram, whatever it is you want to use. Snapchat. Snapchat. And, and, and then tweet out your favorites. Tweet out your favorites. Your favorite thing that you heard, but make sure that you do it to me at Angel Louis TS. No. Just kidding. No, no, no. And here's the question. You get one to three words. That's all you have. I'm starting with Monty should be very easy for you. Don't worry. One to three words. Describe the future of open stack in one to three words. Globally ubiquitous. Jesse. I don't like this game. You can phone a friend. Would you like to phone a friend? How about you think about it? I'm going to think. Mo. Invisible. Oh, invisible. Very nice. So ubiquitous. Similar to. Transformative. Transformative. Pain free. My word is awesome sauce. I'm just saying. Thank you very much for attending. Thank you to our panelists. Thank you folks for taking the time. Thank you for coming. And we will see you all. Thank you. Bye bye.