 It's working. OK, welcome, everyone. So I will first start by introduce myself. I'm Antoine Serois. I work for Contran. I will talk today about what we do for edge computing, mainly. So this. OK, who is Contran first? Contran is a global company. We are presented in 12 different countries worldwide. We have a lot of different facilities around the world. We are a German company. We have facilities in Germany, but also in Toulon, San Diego, Montreal. I came from Montreal. All the communication business units are done in Montreal. We take care of developing all the things for communication and cloud. So what are the typical customers we have at Contran? Typically, we do a lot of OEM design for a lot of telco. So if you have ever used a cell phone, you most likely have ever used one of our hardware. But most of the time it was not brand Contran. It was brands from some big temps. We also have done a lot of different OEM deal for a lot of security appliance, like a DDoS attack or stuff like that. We also sold a lot of those server in the media market, mainly for video transcoding for all TV and broadcasting. So then what we do for edge and cloud computing? So everybody talk about edge. Nobody know exactly where it is. I think it's funny because, yeah, it's all about location. My point on that is it's about location, but we provide server. You cannot put all the same server everywhere. So we have different product for different space in the network, especially for the central office, point of presence, or the access, like street cabinet or cell site tower. What is the main difference you need to look for when you choose your hardware for your edge? First, the physical constraint. The physical constraint is not the same everywhere. So specifically for the edge, you have less space. You have most of the time less power. Most of the time, this is all by telco or in a telco infrastructure. So the telco requirements are the same. You have different requirements for physical security. Next thing is important is really the flexibility. In the data center, you have a lot of space to put a lot of different specialized hardware, like big compute, big storage, big GPU. In edge, it's not the same thing. You are restricted. So you would like to have a flexible hardware is able to do all that in the same manner. Of course, the cost is still important, what you have in CAPEX and the OPEX cost. But the old goal of that is to be able to operate that like you operate any cloud, like with OpenStack or Kubernetes. What is the different hardware we have? So the flagship product we have is this one. We call it SimCloud. What is the SimCloud? This is the front view. So in the front, you have two top of the rack switch. So on the top, there. So this is exactly like having your two top of the rack in standard rack. You have fan and faceplate. All that is in the back. So in the back, you have nine different nodes. So there, you could add a lot of different server. So we have a server like those one where you have two server in nodes. So you could end up having 18 server on that. So it's really good for edge when you don't have space because you end up in that to having 18 different server plus two top of the rack switch. Basically, we take the old rack and we have shrink that in to you. Of course, if you need more, you could pile up that. So different example of module you could have with us. So the first one, the 8040 series is standard one server. And you could add PCI Express or storage if you like. And the next two one is really for compute density. So you could have two independent server by node. So then why this is really good for edge? Instead of having a big pile, you end up to having only few server. In that drawing, I have exactly the same number of compute, left and right. So you are able to really pick and zip that. This also help you to simplify all your wiring or you deploy that. Another thing really important when you try to have many, many locations. We have run couple of benchmarks to explain exactly how we could save power. So in that example, if you compare to standard commodity hardware, we are able to save the power consumption by four. This is mainly because you use very power efficient CPU and very power efficient power supply. And for the rack space is like 4.5 times the saving in a rack space. I talk a little bit about a cost. So the total cost of ownership is really important. So this mean we include on that mainly two things, the cost of acquisition of server. We compare this price with this price. And the cost of operating the server in energy and electricity, this is based on the standard rate in the US. So again, because of the efficiency of the power supply, you end up having less cost of operation. So at the end, you save money by using that instead to using any standard commodity server. If you like to go further in the edge, because this is kind of central office point of present type of server, we also have smaller server could fit like in a street cabinet or in a cell tower. They have more or less the same characteristic, but they fit further to the edge. So it could be interesting. We already have that in production, and we sold that to a lot of people, do a radio access network. So now what we do with that for OpenStack are it's integrated for helping to develop a cloud application. We have done a lot of deployment script in order to deploy on this platform, a full OpenStack platform. So in that example, we have used MassJuJu with Canonical. We deploy three controller, we deploy two redundant node. So with the first six server, you end up having a completely redundant OpenStack distribution. You could remove anything, and you are still able to operate and rebuild everything. You have a place for three order place to add VM. So this means you could have six server there into you. If you need to grow more, of course, you could have more of those server or more of server you already have to that platform. So it's a nice way to be able to deploy very easily on edge location, a full OpenStack controller. This could also be used to very easily deploy on-premises private cloud because it's really easy to install. We do the similar thing also with Kubernetes. We also have developed deployment script to make sure all of that is working properly. So with Kubernetes, the recipe is a little bit different. We just need to have three redundant controller. It's the same idea. You end up into you to having your own infrastructure easy to deploy. Of course, Contran provide the hardware. We help you with the infrastructure for making sure it's working fine. But we don't provide application or software like that. But we validate with a lot of different software, mainly what we call infrastructure and orchestration, all major distribution of OpenStack operating system, OP and FV. We deploy all of those things on the hardware in the lab just to make sure everything is going fine and make sure when you will do your own deployment, that will be easy. We are also involved in the networking because we provide at the same time the compute and the networking. So make sure it's working fine with OVS, DPDK, OpenDelight, FDIO. Of course, it's an OpenStack summit. It's all about the community. So we are also involved in the community. Mainly we work in the edge computing group. We are active in that community. We provide a lot of guidance on what we think could be the hardware to help to define that for a different place in the edge, providing what we think all the company needs for mainly 5G infrastructure. We are a hardware company, but we have made different contributions to different open source projects like Rally or some Canonical Charm. We send that to the community for making sure it still remains open for everyone. Our goal is to make sure it's working fine for everybody. Sorry, it was my last slide. So if you have any question, I guess it will be the time now. No, thank you very much.