 Cool. So we're going to talk about Pluribus Networks and OpenStack integration. My name is Prashant, and I am a principal engineer working on SDN for the last couple of years at Pluribus. For those of you who are not familiar with Pluribus, basically, we are a stealth mode startup three years, funded by some of the big players in the VC industry, NEA, Menlo Ventures, MDV. We have a product which is built on network operating system, virtualization, and we have a high-performance hardware software integration. So deployments in 19 countries right now, we still haven't achieved the product completely. What's our mission? So we are trying to bring in the same economics of what server-hypervisor industry went through in the last decade or so. So what are we building, essentially? It's an OS-independent platform. We are compute virtualization stack-independent. We are merchant silicon agnostic, so we work with Broadcom and Intel chips. And it can run on your hardware of your choice. You can take the OS and run it on white boxes. So what's our product line? Essentially, it's a 2U. We have two product lines. One is a 2U form factor for 64 10-gigipodes with two core Intel sandwich processor. So that gives you a lot of compute power on the switch. We have a lot of storage in the switch for your, if you're doing some heavy-duty analytics, and SSD and hard disk on the box for logging and analytics. Besides that, we have a NPU in our 2U form factor, and then we have a 1U form factor with certain reduced hardware specs. So what's our basic unit of management? On the switch, it's what we call a VNet, so a virtualized network. A virtualized network in a basic form is essentially some kind of an isolation based on VX line or VLine or whatever the customer chooses. But then on top of it, we add on a lot of network services out of the box. We do IP pool management, and we also provide a per VNet storage space from an open-stack perspective. If you're thinking of scenarios where you're going to use glands or if you want to store volumes from a storage perspective, that's a very interesting scenario. And the other thing to note is that we work with different control planes. So things like OpenStack is one control plane for us. You could use CloudStack. You could use vSphere, OpenFlow, depending on what your need is. So I'm going to be doing a live demo here about our plugin. So essentially, this is a quick dive into what the setup looks like. So we have a two-note compute cluster with the controller, and then that's going to talk with a quantum network controller on our switch. Some of the things, we've worked on Falsum, Grisly, going to Neutron now, and then I'm going to be demoing a bunch of these DHCP, and then per tenant features, you know, load balancer, routers, et cetera. All right, so let's dive into the demo then. So the thing what I'm going to do, start with, is basically I'm going to create a network. So let's go in here, spin up a network, right? So let me go in here, create a network. I'm just going to call it my www.webfarm. I'm going to assign a subnet. And then I'm going to say it's a 162.150.0 slash 24. All right, so once I spin this off, let's go back to the switch and see what happens, right? Our switch, this is our switch CLI. So a switch is basically part of a fabric. You could scale up your fabric horizontally with multiple switches. For example, this is a single node fabric here. So if I do a fabric node show, what this shows is that I have a single node in the fabric which is connected. The next thing to note is that we could have multiple services running on the switch. So these show up, once the VNet virtual network is provisioned, what we are actually doing is we are provisioning a container on the backend which will host network services. And you can plug in additional network services using Quantum or something else. So for example, this is our VNet show. So this is showing all the virtual networks on which we are provisioned using OpenStack. And then we have DHCP which is automatically provisioned. So we have a DHCP which is attached to the service. So let's, and then the other thing to note is that IP pools show, right? So IP pool basically shows you the IP pools which are assigned to each of your networks. So right now let's go back and do a network topology. These are still disconnected networks. So what I'm going to do is I'm gonna spin up some instances. So I'm gonna create a web server here. I'm just gonna call it web. And then I'm gonna instantiate four web servers in my network. And then I'm gonna attach it to the WW network. So what this is actually doing is that the VMs are getting instantiated using NOVA. And then once the instances come up, they're talking to the DHCP servers which have already provisioned. So if you go back here, and then if I do a DHCP host show, so I see that the switch picked up and it's actually showing some visibility into what is happening on a per-wee net basis. So the other thing I'm gonna do, I'm gonna launch another instance which is gonna serve as my client. So if I go back here and then I'm gonna say attach to my external network. So these two are gonna be on two different subnets. So if I do a launch here, so and then if you do a DHCP host show, we see that there is a new host which comes up on a 120.2. So now let's set up a V-Router. So if you look at the network topology, we still have two networks which are not connected. So what we're going to do is we are gonna plumb a V-Router in a logical router using your typical open stack workflow. So let's go to router here. I'm just gonna create a router and then I'm gonna add two legs into it. I want to connect my WWO web farm and I want to create my external network. So what I'm going to do here is I'm going to add my WWO subnet. I'm gonna do an add interface and then once that comes back up, I'm gonna set the external gateway. So this external gateway here is going to point to my EXT network, right? So once this come up, what we're actually doing at the back end is that we are provisioning a V-Router in a particular V-Net. So if you go back to the switch CLI and if I do a V-Router show, so I have two routers which are connected to each of the subnets, each of the networks. And then the other thing you notice is that we automatically are running RIP in the background using Zebra and the Kaga suites and they have picked up the subnets. So this is my network topology now. So let's try it out. So we have logically plumbed everything together. Let's go to the VMs and then try connecting across subnets. So I'm gonna use my client and then I'm gonna log into the console. Let me try my wired network here. So, okay. So I'm in my, if I do if config zero, you notice that it's on my 150 subnet, right? So let me try to ping into my, the other subnet I just created, which is 200.1, for example, or let's do a 150.1, okay? So I can connect to my subnet out here. So, if I do if config zero, so I can connect to my gateway out here, 192.168.120.1, right? So essentially we can talk across something is going on in the network. So essentially what we want to do is we can provision your V router or your VLB or your load balancers on the fly on this particular attach to particular Vnet. So if you go back to one of the value props we have is essentially we can create a, using a NetWiser platform, you can plumb in multiple switches and then scale across horizontally. And then each of these, you can run multiple control planes. One is of course you can have your Vnet, which is basic isolation. And then you can have your services zone where you're actually running your, some of your V router and your VLB services. And then you could have your native UNIX applications and then run it on the switch. And also you can run your hypervisor based applications using KVM support, which is already plumbed into the platform. And you can do some advanced services like your traffic class management, your cause or quality of service, and you can do some advanced real time analytics and monitoring. You know, this speaks to the same thing, which is you can horizontally scale your logical network. You can provision your Vnets across multiple switches or top of the rack switches. These are some of the partners we are working with, you know, in different spaces, Red Hat, Tipco, Oracle. And then, so this also serves as an ideal since we have a compute built into the switch. It also serves as a, you can use it as a head controller node to control your entire OpenStack rack. This is just a quick take on where we are in terms of competition, what our advantages are, compared to some of the other players in the same space. So thanks. So we are at booth C8, where we are running a live demo of this whole, pretty much the virtual network services. We'd love to talk to you if you have any particular scenarios you're interested in. So thanks a lot.