 Hi there, everybody. Ooh, loud. I realize I'm between break. So what I'm going to do is I'm going to talk quickly through my slides and leave some time for questions, kind of wake you back up before you go out and re-caffeinate. One of the things, though, from sitting behind the magic curtain here, is as Justin got up and started talking about Uber's evolution into open networking, the amount of excitement and downright giddiness sitting around the table between Gavin and Arpit about, hey, we've been working on this for a long time. And look at the cohesion, how these things came together and how we're able to talk about a common set of systems and a common set of mental models and really the similarities. That's because of open networking. I also just a little side note, when Bob got up and talked about ARM, my physics background is in generating power from temperature gradients. So that PhD was years ago, and now ARM has IP in it. So these things really do come. You labor, and you work hard, and then you get up on a stage and you hear that all of these things are actually dropping in. And so I have a slightly different background than our previous speakers. I'm actually a network consumer. So I work really closely with the AWS networking team who is here. So if you have deep networking questions, go talk to them. All right, there's some hands wave, apparently they're sitting over there. But I'm gonna talk to you a little bit about what IoT needs from the network and what IoT provides into the network. Just for those who haven't spent day in and day out in the IoT space, it's here. It's been here for a while. This is one week of customer meetings for me. So yes, that's facial recognition on pigs to figure out exactly when to feed, when to vaccinate, who to breed. These are important things. Manufacturing, automotive manufacturing. We've got shipping containers with perishables. One of those containers full of crab meat is worth $750,000 you wanna know whether it was spoiled before it went in or not. We've got last mile delivery with drones. So cool. And then of course, flow through airports. Turns out you can basically pay for a smart lighting infrastructure just by solving the problem of where are the wheelchairs in an airport, right? Coolest job ever, like super cool. I get to learn about all these things. But one of the downsides of using wonderful acronyms like IoT or not, I thought I was clever, is that you forget what the underlying meaning of the actual name is. So we hear all the time about IoT, but we don't always talk about the internet of things anymore with customers. Or if we do, we talk about the things. Internet is the operative piece. There have been things for a very long time. They just haven't been connected in the numbers at the scale and driving the same value that they are today. So, so what? Well, the internet, although originally designed by physicists to talk and communicate with each other, really developed into, with a lot of the web technologies, a very extensive, powerful content delivery network. How do we get content out to the masses of endpoints that wanna be able to consume that? And by all means, there's data coming the other direction too. But if you look at the way that most of the peering relationships are built, if you look at the way the mapping and the management of traffic flow is built, it's really built to deliver and cash at the edge. IoT reverses that. All of a sudden, you've got potentially 130,000 sensors sitting on an old refinery sending data to some central vibration sensor management system that's doing predictive analytics on every compressor within that refinery. That's a lot of data moving in the other direction. That's caused some challenges. As companies think through, how do I, how do I adjust? I still have, you know, all of my content delivery to do. We're AWS, we're Amazon, we do a lot of that. But now I also need to look at load balancing in the other direction. The mobility networks. IoT is a blessing and a curse. First machine to machine looked like the best thing ever. I can resell my 2G, 3G networks. They're brand new again. That's like, you know, free money floating underneath. Then came those pesky little startups with their LP WAN and their LORA and their, well, gotta figure out how we answer that. So we've got LTDM, we've got NBIOT, see acronyms, narrow band IoT. All of that has created this shouting match a little bit within the IoT space for I can deliver your data faster, better, cheaper, and securely. There's been a lot of naysaying over who does what, when, where, and how. But there are a couple of fundamental pieces about IoT. One, there are a lot of global companies who have assets in the middle and nowhere. So it's not typically where cellular networks have had their strongest backbones. Populations, all of us with our mobile phones, that tends to make for a better network, for more productive network. Knowledge changing a little bit. Will we get 5G with a different, with a different map, maybe? We still have to use the same infrastructure. But again, IoT, kind of the little thorn, but with big promise, 50 billion devices. Who doesn't want that sitting on the network? And then finally, corporate networks get a hold of mobile devices. They finally figured out they've got their MDM solutions, they figured out the security paradigms, and along comes the corporate fitness tracker, right? HR loves it, they get to learn, they get to reduce their costs potentially within our insurance. But now there's a whole another group of devices that don't play by the same rules, don't have the homogeneity of application infrastructure, of operating systems. So a whole new world that has to be kind of shoehorned into the MDM world. If you're in the industrial space, which even most of the consumer electronics companies who build their own are, you're also trying to figure out how do I deal with air gap networks? It used to be that I could moat off all of my critical infrastructure, and that was my security. And then Stuxnet. And every once in a while, ooh, maybe I need two moats. Okay, well that doesn't work when to take advantage of all of the IoT connected $5 sensors, right? Behind the moat, you get the $50,000 vibration sensor. If you build a bridge over that moat, you can spend $5, and you get better data density because you can put more $5 sensors on than one big $50,000 one. So we've seen a lot of customers struggling with how do I take my corporate network and then how do I figure out how to connect that to the devices? And then where I don't have, where I'm using cloud services, because I don't have the toolbox, I don't have the flexibility of toolbox, I also don't want my engineers figuring out how to optimize the underpinnings of ML models. I want Amazon to handle that so that my guys are focused on actually building those models and running those models. I want them delivering the actual value, not the infrastructure. It's been interesting, let's just say that. It's been interesting. And there are a lot of requests for how we fix that. The cloud brings all these things together. Again, our networking team can talk to you a little bit about what we're doing in that area. But IoT for AWS and everyone else has redefined what edge is. Yes, there's the edge to the network. Funniest thing in the world is to be in a naming meeting where the debate about what is the edge happens. And you've got people have been in cloud for years and they're talking about the edge of the network and then they're talking about the edge of the cloud. And then you've got, oh, but I don't know how, tea. I mean, the gateway is the edge. Where you've got the endpoint, you've got your sensor device, but also AWS Greengrass was a nice compromise. It's our edge product. We didn't have to get too fancy with the naming there. But all of this really moves from a management structure, from a flexibility and capacity structure, from a consistency of tooling and transparency and visibility really changes the way that we as a cloud company experience network. So how has IoT helped solve some of these problems? IoT isn't just a problem. We helped solve a few things too. And we work very closely with our networking teams within the IoT service delivery teams on a couple of key areas. One is just, again, based on the heterogeneity of IoT devices, over the year updates have had to become something that we got really good at, that we understand how to do in an intelligent manner. That's one of those technology stacks that we've been working with our networking guys to figure out how can we leverage what we've built in the IoT space so that the networking team doesn't have to maintain a whole separate stack. Remote device management kind of falls into that as well. Again, being able to leverage that underpinnings of what we've done in the IoT space and that we've been able to turn into kind of a crank the handle type service. Time series analysis. I ran into one of my neighbors. And he's like, hey, I used your new IoT analytics service. I'm like, oh, that's awesome. What are you doing with it? Well, I'm in the networking team. So turns out you guys pulled together a lot of the components that we needed, from time series to data ingest and pipeline preparation to being able to set up scheduled queries, having sort of a one-stop shop to pull all of that together into a dashboard that's fairly simple and automatically updates for daily and weekly meetings. I'm like, ha, ha, ha. Look at that. Network is full of devices. I'm gonna talk a little bit more about security on the next slide, but security is another one of these kind of two-way streets. Security is both top concern within IoT customers and also something that AWS takes as our top priority. So it's something where we've really seen a lot of give and take where IoT is a canary in the coal mine for certain types of threat landscapes and where IoT can also leverage and benefit from network security and frankly, particularly the metadata within the network. IoT does not necessarily carry context with it. So you get a workload. In fact, a little scenario here. We've got nuclear power plant, 130,000 sensors, let's say 75,000 of them start going off. Do we have a DDOS attack? Do we have a spoof system? Or are we in the situation where we actually have a meltdown? There's really no way from the message structures to tell the difference. So we need to be able to have the context around what that use case is and the network is what helps give us that. It helps look at the behavioral patterns and it helps us to determine whether this is in fact legitimate traffic. So working very closely on being able to build that context really starting at the network itself. We've heard a lot about interoperability. Bill talked about information modeling. Actually, so did Justin, he mentioned that was a problem. This is one of those areas where we in IoT have not fixed it and are hoping that you're gonna lead the way. And we work very closely on trying to figure out what is the standardization around modeling structures and how do we take advantage of what's gone before? Because again, when you have the types of function variety that comes with that heterogeneity, find fish is not a common function, but it is a function that exists on IoT devices. How do we leverage what's gone before? How do we understand? How do we use things like ONAT? How do we understand what that means from an IoT standpoint? Are there mental models we should be adapting or is there actually technology that we can underline and adapt, especially when you're combining the function virtualization with orchestration? Those are two key areas where networking has gone before and where we are kind of catching up in the IoT space and seeing what is extensible and where we can help push it. This last one, kind of key too. Open standards, open source, open exchange. AWS is working harder and harder in this area? We do a lot, we don't always talk about it. We're trying to get a little bit better about that too. But in the IoT space, that heterogeneity can be a killer unless we figure out what pieces of it are really standard components that we need to open up and communicate and how do we deliver value out of that? How do we validate that value across the supply chain within the IoT space? It's something that networking has not 100% figured out but is doing much better than we are. So I talked fast, but not quite fast enough. We still have four to five minutes for questions if anyone has questions or not. Maybe we've got break. All right, well, I will let you guys go. I won't ask you questions. That was going to be the next thing, but given that we're already running a little bit over. If you do want to talk to the AWS networking team, they are here, they are the team that builds the networks within AWS or they're an infrastructure team versus a software services team. And they've got a booth outside, so thank you. Thank you.