 Mae'r bwysiglach yn gwahanol addysg ar y cael eu bwysiglach, gyda'i adeiladau o'r gael ddechrau. Mae'r byw y gallwn y cyfathion yn ymgyrchol gyda'i gael. Rydyn ni'n ymweld yo. Rwy'n meddwl i'n meddwl, a gofio'r cyfrifio'r bwysiglach yma. Rydych chi'n ymgeisio'r clwiawn yma. Rydych chi'n ymgeisio'r ddweud o'r cyfrifio'r londig a'r adeiladau'r ymgeisio. Clown computing is not, contrary to popular perception, a new concept. It describes a highly disruptive transition of IT from a product to a service world. And we really don't have a great deal of choice about whether we're going to adopt it or not. Going back to the example of the Industrial Revolution, or what I like to call Mrs Muggan's Inc. I mean, we can know more avoid clown computing than Mrs Muggan's could avoid the Industrial Revolution. So that's the theory. What about the practice? Well, as I said, activities which are ubiquitous will find it becoming suitable for service provision. And that's not just one particular aspect of IT. Those activities can occur anywhere in the computing stack. This, unfortunately, has given rise to what we generally call the different asses of clown computing. So, for example, you have software as a service, or SaaS, where you have providers like Salesforce. You have SaaS, or platform as a service, providers like Windows Azure. And you have EAS, which is infrastructure as a service providers like Amazon. I must admit, I'm not a big fan of these terms, because you may as well just call it software as a product. Platform as a product, infrastructure as a product in the old world, so SaaS, Pat, and EAM. Simply all of this is doing is reflecting the fact that these activities have shifted from a product to a service world. But it has one important distinction. It creates a boundary, a division of responsibility between you and a provider. So, for example, let's pretend that you aren't using a company like Salesforce. Well, you're obviously concerned about your data, whereas Salesforce is concerned about providing the application, and all the things underneath that, so the framework is built in the operating system, the machinery at London. Any one of these layers, whether it's software, whether it's platform, whether it's infrastructure, is all about some components being on your side that you're concerned about, and some components being on the provider side, and what they're concerned about. But unfortunately, this is actually a delusion. The reason why it's a delusion is because your data actually doesn't exist outside of this. It actually exists on their system, and that actually creates a risk. So, the types of risks it creates, and there are two basic forms. There are transitional risks related to the formation of this new industry, and that's things like management, trust where you trust the vendors, where there's transparency in their operations, and the security of supply. But it also creates the normal outsourcing risks of suitability of the activity, pricing competition, locking, and second sourcing. Now, second sourcing isn't important, so I'm just going to pick on this one for a moment. If we take the example that you're using as software as a service provider, so they are concerned with the underlying systems and your data exist on that system, if something goes wrong, and that it could be a disaster, it could be it was bought out by another company, and that has happened, and then closed down, then you've got to somehow get your data out of that system. But the thing is, you get your data out of the system, and that's pretty useless, because all you've got is data. So, your choices are to rebuild everything that they've done in order to run it, or you need to find another provider. So, you shift your data over to another provider, but of course your data's only going to work in another provider, if they're pretty much running exactly the same systems as was originally in the first provider. In order to get second sourcing, you actually need these things, you need portability of data code, whatever happens to be in that provider, you need a choice in destinations, you need interoperability or standard outputs between the providers, and you need easy switching. Without these, you do not have second sourcing. But we've actually been here before. The whole thing about standard outputs, easy switching, all goes back to Paul Brand's work on the security of supply back in the 1960s. Now, for those of you who don't know, Paul Brand's work led to an awful lot of different network protocols. What we ended up with many years ago was a mass of different protocols, IPX, SPX, Banym Vines, Abortor, SNA, Depnet, until we got consolidation in the industry around one particular protocol, TCPIP, principally because it was open, and also because it was supported by upstream open source systems. So, if we just take the infrastructure layer of the computing stack, where we are today is something like this. We have a mass of different protocols and different APIs, none of which have become the de facto standard. We are in that sort of IPX, SPX versus TCPIP versus Depnet stage. Now, EarthVy is one, and that's the Amazon EC2 API, is the most dominant. So, as in Ubuntu, what we've actually done is we've said, well, rather than create our own API, which seems to be the flavour of the month at the moment, everybody is creating either an API or an API of APIs, we would actually adopt someone else's API. So, we found an open source system called Eucalyptus, which actually matches the Amazon EC2 API. So, we introduced this last April into Ubuntu distribution. And principally, this enables all the Ubuntu users in the world to build their own private clouds, which match the Amazon EC2 API. And also, to run Ubuntu on Amazon EC2, so you have much easier switching between the two environments. Now, this is a first step. What we're actually trying to do is create a competitive marketplace of different providers, just like you have within the electricity industry. But in order to do that, we need to get standardisation. We need to get standardisation around APIs. And the reality is that's only going to happen if those standards are not specifications, but actually defined by running code or open source records models. So, to quickly recap, cloud computing, underneath all the noise, it really just describes a destructive shift of IT activities from a product to a service world. And it's governed by a number of different factors, the concept suitability technology and attitude. Now, we can't really avoid this change. I know people talk about, well, we may not adopt it, but you can no more avoid it than people could avoid the Industrial Revolution. There are obviously risks. Some are transitional, will be solved. Some are all to do without sourcing and things like locking and second sourcing options. But there are solutions to those risks, and we've also been here before. So the real question is whether we're going to learn the lessons of history or whether we're just going to actually repeat them. And unfortunately, at this moment in time, in the cloud space, repeating the same lessons of the past seems to be the way we are going. So, thank you. Can I, just before a despair of it, did you all follow that? Did I lose you? A horrified face, though. Did I lose you? We're going to take questions now. We'll take questions up.