 So the next 30 minutes or so, we're going to be discussing, I guess, absorbing and reacting to Max keynote and also obviously welcoming questions from the floor. But before we start, maybe Rich and Steve, you can just give us the 30 second overview on who you are and what you do, Rich. Thank you very much, Duncan. Rich Miller, I run a consultancy and holding company called Telematica, but that has brought me to serial offense many times in Silicon Valley in startups, starting much longer ago than I care to admit. But most recently, in starting with grid computing and moving into cloud, significant high-performance compute, data grids and the like, moving more and more into automation of cloud infrastructure and now into big data and APIs. Most recently, I was the CEO of Streetlight Data and I'm now active in a number of efforts on behalf of large customers in big data, data management and also in infrastructure brokerage. Thank you, Steve. Sounds good. My business is defined by Max Slide 9, I think it was, with the multiple different stocks that you got, Mike. So I'm the CTO of Canopy Cloud. We're called the Open Cloud Company and the first people think, when they see that is, it must be all open source, right? Well, what we really do is we open up clouds and if you look at the cloud business today, it's a fragmented, distributed, non-coherent platform, right? And if you think about people who want to really spend money and solve business problems, I'll give you an example that we've just done a solution with a British broadcasting company who did some recent Olympic kind of events and how do they take advantage of cloud? How do they distribute media around the cloud? So we're helping them solve the problems, but when they look at the cloud market, they see proprietary, they see open source. If you talk to them about open stack, they're like, I don't want to hear about that, Steve. So we're trying to solve that problem with them and that's what I spend a lot of my time doing. I meet a lot of passionate people like Rich and like Duncan and so we have, it was a great slide, it was slide nine, it had the different stacks, the different interfaces, the different technologies. How do you make sense of that? I mean, Simon Wardley, who's now a CSC, you know, he talks about how do we get this to commodity? It's a very challenging thing to do, right? So I spend all my time trying to work that problem out. When you're not tweeting? When I'm not tweeting. So speaking of which, I mean, yes, you made that point on Twitter very recently about, you know, open is not open source. I mean, let me start with, you know, you've heard Max vision for, you know, next generation cloud platforms plural. When you think of next generation cloud platforms, Rich, what's top of mind for you? Actually, what lives on top of them and the first two slides of Max presentation absolutely hit it. First APIs, but more than anything, the pragmatics of managing the APIs. In most cases, the companies I'm working with are both consumers of API for their raw material and then the producers, the publishers. The world is full of bad APIs. I mean, let's let's get that out straight away, right? There's no doubt about it. And what that means is the pragmatics of API management often is to take into account the fact that the quality of some of the APIs you have are really low. I think that Max was the point you made that you're only as good as your API, right? That's right. And, you know, you know, I hate to belittle, you know, keep belaboring the point around what the four square guy said. But, you know, someone that you don't know couldn't use your service if you had a crummy API, right? And they wouldn't want to use your service because you wouldn't have a way of managing the lifecycle around that. And they certainly couldn't extend what you did kind of as a core into a new service that you'd never, the original person never envisioned if they didn't have a solid, easy, consumable way. Because if somebody has to work really hard and has to have a lot of skill in order to use it, it's a bad API. But, you know, Max, I think that's one of the problems I certainly see with, certainly, European customers is that giving up control. Right, so we have a lot of people that they want to control the whole IT experience. And that concept, you open the door, you expose an API, and you let someone do something with your technology that they don't know about. That's a very difficult thing for, I don't know if it's a cultural thing? Well, that was what I was saying before, you know, that's why I call it gene therapy. You alter the DNA because in the DNA, the people that had control, you know, giving up and changing behavior, you know, a lot of things are possible today on cloud that a lot of people don't think are possible. It's because the human behavior hasn't caught up to where the technology is. So, continuing on the theme of APIs, I mean, you know, we've had APIs for years. I mean, I worked on something called FIX, Financial Information Exchange. Many, many, many years. Yeah, I know. But, I mean, what would you say about APIs and scale? Because, you know, a lot of times people talk about self-service. To me, self-service seems anachronistic if one's trying to build platforms at scale. Steve. So, good topic, right? So, I was on the phone with George Rees of Instratius last night. I mean, for me, he's the king of APIs, right? I mean, he's been through all the wars of connecting things together. We talked about three particular things about APIs. One was the security of them. One was there are so many of them. And the third one's about how bad they are. And George, by the way, will always let you know how bad your API is. He's very passionate about them, right? He's very vocal. He's very, very vocal. But I think, you know, from a business point of view, the APIs are great. I just think like, you know, especially in Europe, I mean, my company covers North America, Europe, and Asia. And I see three different attitudes to that whole cloud piece, right? In North America, I think they, if there's any Americans there, I apologize, but they seem to be very malleable to the future. Well, like Mac and Rich for example. Europeans are much less malleable to the future. So, for example, you know, we do software as a service. We offer different applications. We do this thing called Canopy Compose, which lets you create application blueprints and wrap them around with APIs. And it's too new for a lot of companies, right? So, when you say compose, I mean, he's saying that you can create sort of a catalog of these things. Yeah, so there's three bits. I mean, you can pick things off a shelf, you know, web server, you know, everyone's doing that right. Then we can give you some opportunities to blend your own and then write your own completely. You know, if you're a really hip developer guy, we'll give you all these tools to do it. I don't see many of those guys in business willing to spend money right now. So, if I'm doing that, then am I tied into your view of cloud? Or can I take that blueprint and run it on software? Absolutely. I mean, that's a big tenet that we have that you should be able to distribute applications wherever you need, right? But then that brings up a different kind of cloud friction. So, if you let somebody deploy applications in data wherever they want, their business leaders are like, oh, hang on a minute. So you're going to let somebody deploy my CRM in North America because everybody's scared of America at the moment, aren't they? Is that a good thing or a bad thing? I need this running in Germany. So that sounds like an argument for governance and I mean, Rich, I know that's a topic dear to your heart. The second point that I was going to make was Max centering on data. And again, it's the pragmatics of managing the data knowing very clearly where data is, data lineage, data provenance, keeping track of everything built around these great engines that are, you know, incredibly performant, but the, you know, the blocking and tackling of watching data as soon as it's gathered, managed all the way through its life cycle. This is incredibly important and the platforms on which we build all these applications have to be extremely aware and cognizant of the unfortunate realities of jurisdictions, of geographies, where data can and cannot reside, where it must and must not be processed and in what form of, you know, secure, what forms of security you have to apply to it. If you don't have that incorporated into your platforms, into your offerings, you have basically engendered exactly that kind of insecurity and that kind of doubt that you've just mentioned. So if we have fungibility then, or if we aspire to fungibility and we have security, so how do we square that circle, Max? Well, I think one of the things that people often kind of get wrong with cloud is they think, okay, I've got to be willing to give up control. No? You have to be willing to manage what you make available and how you make it available. So you don't want to give up control because if you give up control completely then you have just the wild west. You have no curation of data. So as you're selling services it becomes kind of a flea market. You know, kind of everything's kind of in there. But you want to be able to not control everything to a point where no one can get access and no one can innovate around it. So there's this balancing act of, okay, how am I going to give up some control that makes sense? It's a give-to-get model. If I give up this control by making this API available to this amount of data, because I think this data could be useful for a wide variety of services that I can't deliver myself but perhaps the ecosystem could. So it's that balancing act that I think is important. We've got to give-to-get model for code. And it's started. We don't yet have a give-to-get model for data itself. And to the degree that we actually can develop the moral equivalent of get and get hub for data with all of the aspects of forking, joining, and that kind of reuse. I think those are the kinds of things that really make a difference. We are starting to see data fall into that category. I mean, if you think about Four Square, I mean the data, locality, in terms of where is this mobile person at this particular point in time? That's a data source, okay? A lot of people think, you know, they don't think in those terms, but mobility, locality, those things are data sources. That's information that someone else could use for their purpose. And we're starting to see, you know, kind of data repositories of different kinds of data which are foundational elements for you to build interesting, more collaborative, more innovative services around that core asset. So we tend to be somewhat anthropomorphic and think that, you know, 10 billion is a big number because that's the number of people and, you know, world data sources and so on and so forth. But there's another kind of internet out there that everyone talks about, the internet of things, everything, I mean, and the scale which you sort of postulated where, you know, it's orders of magnitude greater than that. Steve, does Canopy handle internet of everything? No, not today, we don't. I think the biggest scale of things that we do is around media, so we just built this new media cloud. And the scale of that is frightening. And I think one of the messages that I think Max did really well today is, you open Pandora's box and this thing is crazy, right? So one of the things we learned early on is as soon as you do something cool and everybody wants it and you can't scale, either you're going to be a Google and spend 20, how many of us, is it 21 billion, 23 billion on infrastructure? We don't want to do that. So when you say scale, what axes are we talking volume of data in the media or are we talking processing power? Well, it comes back to old tradition, I don't know how old the general age of the audience is here, but it comes back to a simple thing called the network, right? You try and send all this media stuff over the network and all this chatter, the different types of networks that you have. One of our major investors at Canopy is Atos. They've got a very traditional network. They're very, very good at doing things in a traditional way. You open the door to the new internet of things and media and it's like, well, this is totally different. So for me, open, for example, and when we talk about cloud open, is how do you partner with the guys that are really good at that, which is the Google's, which is the Amazon's, right? So you touched on that, I think, Mac, in terms of the sort of the fiber overlay network and, I mean, just a plug for a company here in Europe doing that inter-route. I've said there only ever be a European company unless they call themselves inter-route. But the point is that it isn't just about clouds of compute. It's about how you wire those up. And I get the sense from looking at software and talking to the engineers there that if I want to, I can run a very secure, highly distributed, global private cloud and never touch the internet unless I choose to. Yeah, so, you know, just taking a look at kind of internet of things. You know, if you don't have the right type of architecture in terms of how you do overlays and how you automate, then you'll end up with the internet of some things, not the internet of things. Because you won't be able to scale it. You won't be able to adapt to the different geographies. You won't be able to ensure that people can get the information that they need at the time that they need it. So this aspect of predictability, I mean, if the network gets congested and all of a sudden it starts dropping packets, what happens is it resimits and it gets more congested, right? So you have to have an approach that gives you the flexibility of doing things over the top and doing things in an overlay fashion. And so, you know, for those familiar with Akamai, this kind of model has existed. The internet wouldn't have been the internet without this type of an overlay type of a model where you do things over the top in a very automated fashion without being kind of bogged down in the physical network. And this aspect of the explosion of mobile and these embedded devices and these back-end cloud services that need to interact with those devices, if you don't have that same architectural model, you're not going to be very successful. We've talked about the asymmetric nature of data and it's great that you mentioned Akamai as a model. If you just looked at the topology, it's exactly the topology you want for the internet of things, except you've changed the polarity. That's right. More than anything, you're not talking about a data, a content distribution network. You're talking about a data collection network. It's rather than CDN, we're talking about DCN. And turning that around, turning that on its head is no mean feat. It's not an easy thing to do and the kinds of things you have to consider, congestion, the fact that data has gravity, has weight, creates friction and then all of the issues that you get into regarding the authentication and security of that remote spot from which you're getting data, it is a completely new set of problems that no one yet has addressed, but they have a number of the same design patterns on which we can depend. So that's a kind of neat segue into something else I wanted to talk to you all about, which was a point you raised around, extreme innovation misses the point if you're not also looking to optimize. So perhaps this is an example where one wants to optimize what we have rather than the somewhat hubristic, everything that's gone before is kind of meaningless, doesn't understand the problem, let's just toss it all in the bin. So, I mean, I thought that was a really interesting point because we talk about hybrid and we're talking technically about hybrid. This was a hybrid notion of, you know, Steve, combining optimization with innovation. I thought that was a really interesting point. Well, it feels, you remember that picture that I think Stephen Hawking drew in his brief history of time where the singularity gets stretched. It feels like that right now, right? So on the one hand, from where I'm sitting, you know, and we've got a, compared to IBM, a very small business, but it's a $10 billion business, and we've got, I would say, 95% of people in the old world and 5% in the new world. And you feel the stretchy thing, right, where they want to keep buying physical service, which is good, but now they want to do it as a service, which is very enough. But then you have to, like, tie in all the billing, the OSS, the BSS, and all the stuff around it. And you just, it's just a massive amount of work to take some of them from the old world to the new world. And then I think this audience is very key because, you know, I get accused of turning up and being the guy that talks about the cool stuff that doesn't make any sense, right, to some of these old European guys. And often there's usually an open source guy, sat next to me, and he said, we can do the other, we can do this other thing, and it's... There was a Twitter storm recently about, you know, OpenStack is going to eat the world, kind of, which is nonsense because OpenStack is important, and it's obviously an innovator. Zero VM. Zero VM's wiki page. I don't know if you saw it recently, but it made my blood boil yesterday. There was three things I hated about it. Very relaxed guy, normally. There was three things I hate about it. The first thought he said is, it's open source, everything's brilliant, right? That gives the open source community a bad name, right? Just having that kind of myopic view. The second thing he said was, it's built for a class of... Hypervisors today are built for applications that don't exist anymore. And I have an issue with that because, you know, six out of seven dollars are spent on the old stuff and not the cloud stuff, according to IDC. And the third thing he said, he could aggregate many physical devices. And I'm like, well, what are you? What are you? Aggregate physical devices? What do you think? Aggregate multiple devices? Yeah, aggregate multiple blades. Is that a cool thing to do these days? Why do I care if I'm an application? I'm still angry if I read your wiki page. Well, the whole point, I think, is, you know, what I wanted is the use of set of abstractions that give me exactly what I want, what I need to play with, and those issues about which I shouldn't have to care and that varies, that abstraction varies, isn't it? Yeah, that's exactly right. And just to kind of pull back the curtain inside of IBM when I started the due diligence for SoftLayer, you know, it was one of the things that was the most difficult inside of IBM selling the fact that SoftLayer allowed you to have a bare metal cloud. Like, a bare metal cloud, you can't have a bare metal. That's not cloud. I'm like, what is it? What is cloud? You have elasticity, you have consumability and you have a pay-as-you-go model, right? And as long as, you know, me as an application provider as a developer, if I have a simple API that I can use that allows me to get to the compute, the storage, and the network services that I need, do I care how you deliver that to me? I shouldn't care. Whether it's OpenStack, CloudStack, VMware, it doesn't matter. As long as it's the right elasticity, the right consumability characteristics, the right quality of experience, the right cost point, I don't care. As long as they can get it to you when you need it. That's right, exactly. I think one of the challenges that I see though is, I mean, I totally agree with that, but I'd still see people's buying patterns and people's understanding of technology. That's right. Human behavior is in the way. I think it was 4.5 research that did something recently and they said the biggest barriers to cloud, 82% of people said they were non-technical. It was people, politics, budgets, all the other stuff that you get. One of the cloud stacks that we've got, and I say cloud with double quotes around it, because it's a little bit outsourcing, it's a little bit self-service, 70% of the distribution of that is physical service. When I became CTO, I said, why is it physical service? Didn't we solve this problem 10 years ago when I was at VMware? People feel more comfortable and they're willing to spend the money to buy physical service. They're willing to spend 30 things and get in the API, it's just familiar to them. I think that's the challenge. I think the eye-opener for IBM around software was bare-metal cloud, meaning you can have it on the same term. I see a huge market for that. I think what's fascinating is, unfortunately the person who wrote this I can't remember who they were, which is probably good for them, but there was a big blog post, oh my God, this bare-metal cloud is brilliant, you get your own resources. Totally, I've never, so obviously this person was born very late in the day. I think we've gone from server-huggers to cloud-huggers now, that's the other thing that's driving me crazy at the moment is people are passionate about their stack. Canopy, the name was picked for a reason, in that we see the stacks as leafs, we see Amazon as a leaf, we see Google as a leaf, we see open stack, we're part of the open Nebula CERN thing, we've got our VMware pieces, we see them as a collection of leaves, and then when we talk to big companies, they're like, how do we make sense of this? How do I make it programmatic? How do I program that cloud of all these distributed leaves? That's a big problem. If we can solve that together, that would be really cool. So we've got seven, thank you, we've got seven or eight minutes left. If people want to join the discussion and ask questions, there are mics at the back, so please walk up to a mic and let's keep the conversation going. So abstractions are great. One can have too many of them, obviously, but if we go back to a point you made about data being the currency, the de facto currency of cloud, what is the lingua franca of cloud? Big question, but how are we going to talk about cloud? Are we going to talk about apps? Are we going to talk about services? Where is it going to be most effective to actually build out this canopy? I have one answer. Okay, one is good. So the one thing that I found works is it's a non-technical answer, I'm afraid. It's when you get someone comes into your office with a business proposition and you help them solve that by helping them compose applications from four different vendors, put it in an app store and help them consume any infrastructure stack wherever it is. In our case it was media cloud. That for me is cloud. So I think, you know, Mike's point about the four square guy, when you open up your API and someone's making stuff out of your stuff that you never even thought about, I think that's amazing, right? In my world we have to help them do that. But I think that for me in cloud you just help people do stuff. I think that's what it really is. If that's not too vague. Yeah, it's almost the definition of success. If you're creating a service and you're the only one that's developing the service and the only one that's innovating around the service sooner or later someone would maybe not as good a service but one that's more consumable and people can add additional services around because you've done a good job on the API managing that. The sustainability of that. I mean, to Duncan's point around OpenStack, I mean, you know we have this danger inside of IBM and other big companies that, you know, it's so hard to create momentum that once it starts everybody kind of rallies around it and then it's the answer to our world hunger and the national debt. Yeah. I think there's a tangible. We do have a question about it. It's just a very, very quick point. I think there's a tangible bit about this. When you run non-cloud operations, you know the old world where you set thresholds and you kind of understand everything that runs on it. I think when you're really running a cloud you suddenly let go and you think, I don't know what's running on this thing. You have to do new tools that understand what's normal. You know what I mean? You know when you're running a cloud. That is the point. Hi. Do you want to just say who you are? Hi. I'm Colin Hicks from Edinburgh University so kind of local and this is actually kind of I'm thinking smaller. I'm thinking about our university and not so much the cloud but your point about APIs. I think it's fair to say that our IT provision in Edinburgh University is kind of federated. There are quite a lot of small units doing their own things and I've kind of been on a sort of personal quest to get people to put APIs into the services they build locally. Good one. And I'd love to have some ammunition to take back. Wow. Like it doesn't work unless you do this or something more forceful like if you don't do this you don't get a point right? What I'd like to do is consume... What's in it for them? I mean that's the classic question, right? I'd like to argue that what's in it for them is that I and other people who are not part of central IS could stitch together some of the services that they consume and build features that we're asking for ourselves rather than have them do it. When you say federated I'll say fragmented, right? It could work for it. Is there 2, 5, 10 or 8,000 different? So there's I think there... Are we in the order of 10 or 15 different? Can you give us a concrete example of the sort of thing you'd like to do which you can't do today? That might help. What I'd like to do is offer for instance the people in our school which is engineering to take some of the I'd like to stitch together some of our local data and data that's held centrally so there's identity management information, for example the university holds on people that is quite high level but we've got some quite a lot of local stuff that's the resource. Well, in a wheelhouse. A couple of things. One of the big things you have to deal with is the sense of is both the sense of safety or both it's the reality of safety and the impression of safety for the people and for the community that you have to sell this to. So you've got a number of issues regarding how do you manage the APIs but more than that how do you deal with everything from managing jurisdictions where data can and can't live and you're going to end up making the case that if I do it on the basis of a common set of exposed APIs APIs that actually do in fact publish a great deal of information about themselves and what's behind them that safety and that impression of safety that feeling of safety is you can allow the constituents to build for themselves without endangering the whole if there is a problem you can localize the problem and it will always be contained. And I think you need to speak to your Vice Chancellor about how you monetize your APIs I think that's a very important point. So the Vice Chancellor is a very interesting guy because he's actually in the vanguard of massive online education. So I think it's already happening it may not be happening in your part of the University but if one can offer Edinburgh classes to the world and then I think it ought to be possible to solve problems within the University as a whole. And one of the things I think it's important to understand is going back to the change in behavior again you don't have to have a way of accomplishing this kind of spontaneous innovation that's outside of the core where you can do 100% of the things that you need to do 100% of the time. If you can do 60% of the things 100% of the time that's really powerful if you think about it. And so we used to have this mindset that we had to have everything completely polished and completely perfect before we made it available. Now you want to make something available that's very easy and allow these people that are outside of the core asset to do things 100% on their own or close to 100% on their own but it doesn't have to be every single use case it's not a sustainable model. So what you have to do is break it down into very consumable services that you want to make available 100% of the time and through a self-service kind of fashion but it doesn't support 100% of the different use cases. It's the art of API design. I'm going to have to just wrapping that up just swing by the tech cube there's a bunch of interesting companies there that would love to talk to you about your problem. We've got less than a minute left so I wanted each of you Steve starting with you working away to Mac who obviously did a fantastic keynote. What's the one single takeaway? I know it's a tough question that you'd like this audience apart from just coming up and talking to you afterwards what's the key takeaway in terms of where do you see the next generation platforms going? The next generation platforms I mean you're either on the train or you're under it right I mean this thing is moving really really quick and I think if you want to be on the train you've got to be application focused and you've got to work out how do I make things that question was great so from the APIs how do I sorry to be rude but how do you make money out of it I mean that's how do you make money out of the cloud that's been on the train. I have to agree the point though is making a decision at what the idea of abstraction you are you're either endangering or enhancing your level of business and this is where the art comes in this is not yet science it's not yet the craft of engineering. I think the main thing is don't be afraid of trying something new but you do it in a way that if it fails it fails quickly you recover quickly from it and you can learn from it and so that adaptability you have to plan for that you have to figure out a way how can I gather information how can I incorporate that learning into the next instantiation and you don't want to spend a lot of time and a lot of energy up front you want to get it out quickly learn from it quickly you either learn that it's massively more successful than you thought or it's a miserable failure and you need to do something else very very valuable to learn very quickly this is a classic Silicon Valley fail fast absolutely on that I'd just like to thank everybody we are out of time so if you'd like to put your hands together and thank the panelists for what I hope was an interesting and engaging conversation and then I think there's another keynote up next thank you guys