 Felly, feli rhoi mewn sffa modd i'r ffordd i gyda'r support bod riesa, dzycho gi Sigurffa yma, i fel dod i Doca. сейчасu hwnnw nid championships mae'n heldiau no i fynd gennych amru, yn y twyr ulith gwkea predecessoru fel y gw�제 pan gwybodaeth hon i ceny de禄 rôl gwmpwr ac arено vois làr betharnau hostau ar ailwchiau ac fel peth yma. Un darn drw及, mae amser byddwn asiadoedd yma ar mawr ochr, i comedigca health yma. Felly, mae un angarod sydd beth yr oversefdau diyardweb New York i регeon New York is on par with our San Francisco summit. They're the two biggest summits and we're about to announce things. So the San Francisco summit in February we announced machine learning and we announced elastic file system. Just last week on Thursday in New York we announced a brand new service. I didn't even know this was coming. I heard about it on Thursday night or Friday morning or whatever. It's called Amazon API Gateway. It's another fully managed service and it enables developers to basically build an API with no servers. This is the kind of evolution of serverless computing. You might have heard about Lambda, which was announced last November when it became generally available in about March. You can start now building apps. First you have servers, then you have virtualisation and then you have virtual servers on your hardware. Then they kind of move to the public cloud and all the hardware is abstracted and you just have virtual machines or instances. Then you have Docker and EC2 container service and it's further abstraction to routed jails or containers for even smaller instances of your compute. Then the next layer of abstraction I think we call it Lambda where you can literally just write your code and you deploy your code to the cloud and you call it and you just pay for iterations of code. Lambda was the first service we released in that space. Amazon API Gateway could be the second service. Amazon API Gateway just roughly showing you where it does because it's only been out a few days and I haven't had a chance to have a good look at it. Your requests come in, you front them with other AWS services such as CloudFront. This is your API gateway. You might be familiar with CloudWatch to monitor it and you can set your maximum number of transactions on your API. You can set an alarm if you hit more than 50 requests per second on your API. Cache API calls, so anything that's a cacheable API call has got a built-in caching mechanism. But here's the interesting bit. It can interface off the Lambda. We probably need a Lambda icon there. So you can call a function in Java which will go and get or put an object in S3 or go and get or put an entry into DynamoDB. The API gateway can interface to any existing Amazon service that we've got. So you basically put an API in front of it but without any service or an existing publicly accessible endpoint. An existing API endpoint, you want to interact with it with new functions and new storage or new DynamoTables. So I haven't done a great experience. I haven't given you a great description because I haven't had a look at it yet. But it's brand new. You can create new APIs with no servers. So it's fully managed, it's scaling, it's reliable, all those kind of things you get from the Amazon services. It's three and a half US dollars per million API calls. So again, pretty cheap. You're doing 10 million API calls, you're paying $35 for a service. You haven't got to make it scalable, you haven't got to make it highly available. That's all done. $35. Not as much as it used to cost to build these things. You do also pay for your data transfer out like you do with a lot of Amazon services. You don't pay for a request in but you pay for data out and it starts off that the first tier is $0.09 per gig of data out per terabyte. So pretty cheap data output, pretty cheap number of calls. So that was the first service I wanted to mention. Next one, another completely new service that is now. It's called AWS Device Farm and it's a service that allows Android and FireOS developers to test their app on a range of hardware devices. So we stock a bunch of devices with a bunch of versions of Android, with a bunch of versions of FireOS. And you simply upload your app on to AWS Device Farm, it runs the test on a whole fleet of these devices. And it will give you, within minutes, an output of bugs, performance issues. And if you've ever written an app for Android, sometimes you'll find some of them have got a slightly different chipset, or you know, the Intel processor, some things just didn't work on just some devices on some versions of Android. And you only find out through testing. And anyone who's running a significant mobile app ends up accumulating piles of apps. And then you need piles of apps with piles of versions of FireOS. So you know what your end user experience is like, as Joseph was saying. You really need to know what the end user experience is like. And without testing it, you don't know. So device testing has become important for mobile developers. So a whole new service to enable you to just upload your app, and then it will give you the output of that app run on a load of devices. It does integrate with your continuous integration environment, such as Jenkins. So it can be part of your continuous integration and send it off to the device farm, get your results, it's good, do a deploy or push it out. If no setup, your first 250 device minutes are free. I'm not sure what it costs after that. So three other announcements that came out on Thursday. Service Cattle, which was, again, they talked about it in March. Oh, sorry, this was an app didn't reinvent. It now becomes generally available. It's a new service. You can use it. It's you have an identity access management role, or IAM role you'll be familiar with. And you can set a certain set of services in that particular role, catalogue. So, you know, that group of ops or whoever it might be can use S3. It can do deployment of this stack into dev stage. A different role can do deployment into production. And it's like a service catalogue part of CMDB, but it's become available. It's the most exciting. AWS Code Commit, on the other hand, is a bit more exciting. And that is available from Thursday. So you can use it today. If you go into the AWS console, you'll see Code Commit there. Code Commit is a Git compatible source code repository. So fully encrypted is, and again, it integrates with your identity access management. So if you've got a big AWS environment, you can keep your code within the environment and you can use your existing roles to access sort of Git compatible source code repository. So that's available. Code Commit, the other one is Code Pipeline, which is our continuous integration tool. Code Pipeline will extract out Code Commit. It will run your unit tests. It will integrate with existing CI tools like Jenkins and Hudson. Fire off unit tests. And then the last stage of the CI CD pipeline would be continuous deployment, which is our code deploy product, which is not up there because it's already out. So we've got the full stat now from source code to CI to CD. It's relatively new for us. And the continuous deployment tool, Code Deploy, was actually based on the internal tool that Amazon have been using for years to do sort of 100 to 200 code releases per day, which is called Apollo. It's an internal tool called Apollo. It's obviously AWS EC2 autoscaling aware, so you can do blue-green AB deployments. It's aware of your autoscaling group changing all the time. They use it for the Amazon website. They've been using it for years. And it's far too reliable, releases it with the rollback, and it's now available as Code Deploy, complemented by Code Commit and Code Pipeline. So they all became generally available. So new services in the console for everyone to use. And I will spend just a few more minutes, and then we can socialize outside. And that's the file system was announced. I've got a preview of it on my account, so we'll just have a little look in the browser and I might show you the amount of file system. And I tested the IO on a T2 micro, which is our smallest instance. I do a DD of Dev Zero, and an out file to the NFS file system, and I'm getting 100 megabytes or 80 or 90 megabytes a second, so saturating and geaging the icon of T2 micro to EFS. So it looks like it's pretty fast. It's elastic. You don't have to specify the size of your file system, so if you've ever provisioned a NAS volume, you have to choose the size and then you might be able to grow it, this you just provision, and you pay for what you use. You use one gig, you pay for one gig. And it's 30 cents a gig, if I remember rightly. So if you've got a small dock room, it's pretty handy. You'll be paying 30 cents a month for a gig of stuff, for a share of a huge number of EC2 instances. It's built for scale. So it's EFS, because it's in preview at the moment, but I've got it enabled on my account, so I'll show it to you here. Elastic file system, really good. Really managed file systems for EC2. So there's a few that I've got in there. There's actually not a great deal to it. You just say create file system. I'll put it in my default VPC. I'll have an endpoint in each of the availability zones, and I'll apply a security group to each endpoint. That's name, and I'll call it HG user group. Demo, even. Review and create, there we go. My three availability zones, my security group, my tag, create file system. So it couldn't be easier to create a NAS volume, and there's effectively no cost while there's nothing in it, and you're only paid for the stuff that's being used. So success, I'll be surprised if it's available that quick, but it says available. Now, in terms of my demo, I've got one here with 37 megabytes. I did this actually for the WordPress demo. I've got it installed on here. So there's another one again. Three subnets it's connected to. There's the DNS name. There's a DNS name for each of the endpoints, but more importantly, this really handy EC2 mount instructions. You install NFS util, so it tells you what to do. Make a mount point. Then sudo, mount minus t, mount type, NFS v4, because it's NFS v4. Then this command here, curl minus s of your local loopback metadata, it will give you the availability zone you're in, so you can use that programmatically for each of your availability zones. That's the NFS file system name, and mount on slash EFS. Now, I'll briefly show you the performance if you want to see what it looks like. So my two running instances, both mount my file system, grab my instance name. Okay, connect to that. So there is my mounted EFS in US West 2. It gives you a slightly weird thing for size, because as I said, it's an elastic file system, there is no file system size, so I don't know what it's interpreting in for it. I've got it mounted on this location. So if I go there, and at the same time, I'll just handily save my DD command. DD infiled dev zero, which is a bunch of zero data. Outfile var www.html bigfile block size 1024k count equals 1,000. I'm probably going to start off with a smaller one, but that will write 1,001 megabyte blocks, which I think will agree is probably about a gigabyte of data. And if the EFS is performing, it will give me the time it's taking to write that. There we go. 1,000 records in, 1,000 records out, 79 megabytes a second. So I was quite impressed with that for an EFS file system, 100 megabyte a second. That's like 650 megabits over the wire on a T2 micro. So I'll just have a quick look at it to see what's there. So a couple of other things tonight. Reinvent, this slide's been out of date. It doesn't say to save the date of registration opens in May. Registration's open for reinvent. It's in October the 6th and the 9th. It's the annual developer conference. It's in Las Vegas. Everyone's welcome to go. It's a week-long set of events. Got to get to Las Vegas and it does cost money for the conference, but there's hundreds of sessions at re-invent. It's growing in size. About 10,000 people or something now. This is our AWS user group. It's your user group as much as it is mine. So you want to speak, come and speak, let me know. You want to host to your venue. We could probably host to your venue as well. I think we were speaking to someone about hosting. Stay in touch on the Meetup website. Thank you very much. So we can enjoy a coke bit of soft drinks or there's a beer tap out there. You've got a few minutes we can socialise. So stick around.