 Cyfnodd panethau eu cynnig yw'r iawn i'r ddeithasio ychydig iawn, wedi'i gweithio ar y platforma y cwmaint a'r gweithio ar y gweithio. Yn y cwestiynau, yma yw'r gwaith yma'r gweithio? Yn yw'r gwneud yma'r gweithio ar y Cymru? Yn yw'r gweithio ar y Gweithio? Ac'r gweithio'r gweithio ar y cwmaint ond am yw. Mae hynny'n ddim yn gweithio gan gweithio, ond mae'n gweithio'r brugwad arweithio Mae'r cynllunydd yr oedd dechrau yn ymdweud. Mae'r cynllunydd yn cost. Mae ymddiannau i'r adeiladau yn Ysgol yn dechrau'r stryb yn Ysgol. Yn y gweithiannol, mae'r adeiladau yn gwydydd ei wneud arall gael 16%. Roedd yw'r adeiladau'r syddio yn gyflwydau dyfodol ond bydd ir i gweithio'r gwylliff o'r adeiladau'r gyflym ar â'r adeiladau. Aberth ddod amwys mewn nesaf y dweud i'r cyntaf a'r adeiladau'r cyflwydiannol. For other customers, not being on a US provider may be something applicable. So you may have data that unitized for customers that isn't applicable for safe harbour or the Patriot Act, something you may need to consider. The other thing is availability. AWS this weekend had an outage. The reality is it's all providers have outages at one point. But should you adopt a multi vendor strategy to protect against that? And then finally control. If you stick everything in one provider against one set of tools, if that provider has an issue, your site's going to go down. And it's an important thing to remember that you are the SLAT, your customers, whether those are internal or external. And finally performance. The reason for this talk is why build it on a different cloud. Well, our cloud is noticeably faster than the others. So what is virtual data centre? So first of all it's built on our great big network. Our network is global. It covers all of Europe. It covers right the way across into the states and across into Asia. And on top of that virtual data centre is a huge infrastructure as a service platform. So it's CPU, RAM and storage in locations right the way around the world. It's built into our great big network. And this is one of the reasons why our cloud is a hell of a lot faster than the others. And the interesting thing that this brings to you is as it's built on our network, you get that network completely for free. So if you build a site and it's hosted in one of our zones in London and you also build it in New York and you also build it in Hong Kong, the front-end public access is in each of those locations. But you also then get this private back-end network to do all of your sync replication and backup. It means that all of your private stuff stays private and your public stuff stays public. And that back-end integration is one of the key big performance factors. If you can keep things private and away from the internet, you get a huge performance increase when you need to do back-end services. AWS is your rack space. Every other cloud provider pretty much does cloud like this. Three separate zones all connected over the public internet. If you want to link those zones together, so if you want to link AWS Availability Zones multi-region, you're going to likely do this over IPsec or through SDN, all over the public internet. And for any of you that are devs that aren't network devs, it's often an absolute nightmare or particularly a pain to actually configure. So the main thing is you go for a multi-zonal approach, it's all running over the public internet. You've got no SLA on how it's going to perform. The way that we've done it, the same three zones, but with a back-end on this great big private network. So the guarantees for the interzonal connectivity is absolutely assured. So from your end users on the public internet, right the way through to where the computer is and to where the computer and the different zones are, there is an end-to-end SLA right the way across that. The platformers itself as well, we do straight normal cloud. So this is shared computing on shared servers. All the RAMs guaranteed and you get a two-to-one contention on the CPU. It's still pretty quick when it's compared to the other providers. The thing that then makes us start to be slightly different is we then also do dedicated hardware, so dedicated blades. So one of our reasonable-sized customers, Photobox, who are a big European consumer sales where they do things like coffee cups and mouse mats. These guys in the Christmas run-up, they took 47 individual blades from a 72 terabytes of RAM. They turned it up for an eight-week period and then ripped it right the way back down. So you get performance through doing this on dedicated hardware but as a service. The other thing we do is then do straight-code location and still all tied into this one great, big, private, predictable network. The other thing as well is why is the performance? Where do you have performance from your end customers that are out on the public internet back through to your own site? So one of the main things that causes this is actually latency. So latency for many sites if you're on a provider where everything's over in Dublin and you're here down in Spain, you're 12, 15, 18 milliseconds away from this. It's quite a long way to be if you've actually got quite an interactive website. The closer you move your customers to your actual websites and to your servers, the quicker it is. The performance goes better. And as the performance on each server gets better, you need less of them, so it lowers your cost. So across Europe, you'll see here each of these blue zones is where we've actually got data centres. So it's not the case if we're just in Amsterdam in Dublin. We're here. We're in Spain. We're in France. We're in Italy. We've got two sites over in Switzerland, two over in Germany. All connected onto that private big backbone network, but also far closer to your customers. Over in the States, we've got East, West and Central. And over in Asia, we've got Hong Kong and Singapore as well. So onto the performance bit. What does it actually mean? So the first chart, this is latency. This is us compared with all of the providers. You'll see the latency on us is actually quite a bit less. But the important thing about the performance is is you build cloud into a private network. The public side of it gets a hell of a lot quicker. You're not going to see from the distance here, but on the left-hand side, this is how quick to get over the public internet between us, Amazon, Rackspace and Azure. On us, everyone was running about 500 megabits or less per individual server. We were running about 1.2 gigabits a second. You take that on a consumer level where you're presenting out to the public internet. These things are a hell of a lot quicker. So, finally, why do things differently? Why go to another cloud vendor? So if you build it into a big private network and one that's controlled end-to-end, things are measured really quicker. What we do is we provide fully managed services. So one of the other things that's quite different to AWS and others is we provide a fully managed hosting service as well. We'll do up to the OS. You then look after your stack that sits on top of that. Private by default, if you go global, you build it into a private network. It means that you can start to build out zones and actually get close to your customers but keep back-end control and security built into the network. Finally, location-specific. Don't just host everything in Dublin or one place, especially across Europe. If you've got customers that need things in individual European territories, we can actually certainly help with that. Finally, come down to the stand downstairs. We'll give you a hands-on demo. I think in all the free bags that everyone's got, you've all got a month-free triad to have unlimited resources anywhere in the world. So do definitely give it a go and see how it compares to your existing cloud provider. So any questions at all? No, the question was, do we have anything in China? No, not at the moment. We are speaking to two separate customers. One for in India, one for Australia as well. We generally build extra zones as they come along with the customer requirement. So if there is one there, we can likely build it out pretty quickly. Great. No more questions. Thank you very much for your time.