 All right. Great. Thanks for coming out, everybody. Today, my name's Michael Ford. I'm from MiDakura. We are a software networking overlay, a plugin for Neutron. We also do Kubernetes. We do a lot of container and Docker work. We are founding members of the project Courier for containers. I oversee all of our field SEs. I'm director of technical services. And what I want to show you guys today is our latest product in terms bringing more operational tools into SDN's inside of OpenStack. We have a product that's called Mem Insights that we released. And also another product that's pretty cool that we've been demoing for the first time at this conference, which is Mem Fabric, which I'll talk about. I apologize. My slides are designed by an engineer me. So the last ones were pretty snazzy. So let's get right into it. What is Mem Insights? It's a complete analytics platform built on top of our open-source Meadonet. Meadonet is a completely open-source technology, but the operational tools that we bring are kind of the proprietary bits on the enterprise side. So with Insights, what you get is you get a full health overview of your entire system. And this is super critical, and it's one of the biggest pain points that we hear most when we're talking to operators of OpenStack Clouds or Docker or containers, is that they need a way to see at a very real time the status of their network. We call the features that we call real-time flow tracing is kind of the bread and butter of what people will do in an operational environment. This gives you the ability to see when a packet comes out of VMA on Compute A, goes to Compute B at VM2 over there, normally that process is kind of a black box, and nobody's able to see what happened to that packet. So with real-time flow tracing, you're absolutely able to see that. Insights also has a dedicated flow history back-end that is based on loosely on an elk cluster that we could do storage. And so this is really important when you're talking about compliance, when you're talking about security aspects of a fully operational cloud. You can go back and search flows and history of what VMs were doing at specific times in the past. So with that, I'm going to get right into a demo and I just kind of want to talk about what we're doing with Insights here. Let me come on. There we go. So this is the control panel for Mem Insights. It's a very simplistic view when you're going to log in. And so what we're showing here is our Meadonet Manager 5.0, which is Insights. So when you first log into the platform, the first thing you're going to see is a health status of your entire ecosystem inside your OpenStack. This includes all the hosts that are running our agents, the bridges, the routers created, how many tenants, all the ports total inside of the system. And also, if you look at the bottom right, you're going to see about what's our VTEPs. And so Meadonet itself can act as a OVSDB controller and program top-or-act switches. So that's a pretty important feature to our customers. Now what we're seeing is when you click on the host list, you're going to see every host inside your OpenStack cloud that's running the Meadonet agent. And when you click into the agent, you're going to be able to see the statistics that's happening on that node in real time. So in this case, we're saying, this is the throughput that's coming on this complete physical compute node. So you can go through, take a look. You can see if the status of the agent's up. You can see how much traffic is going across it in real time. So you can also spot performance problems. So if you have a VM that's kind of saturating the entire link, you can see these problems. You can also go back one week. What was this compute node doing for an entire week? Seven days, one day, one hour. You can delve down into it and see it in real time like we're seeing here. So it's really handy when you if you have some complaints from some customers or users, you can kind of click through there. But you can also see agent configuration items. And so now we're moving configuration of the entire platform out of the CLI world into a GUI world where real operators are going to be kind of maintaining this cloud. So here you can see all the different kind of metrics and items that you can change. You can also do stuff with our BGP gateways and changing BGP timers. In the next case, we're going to talk about the routers that we use. So we started at the host level, which is kind of an aggregate view of everything that's happening on that specific compute node. But in this case, we're going to click on an edge router. So this is something, if you created an open stack, you would see all of this inside of the platform. With the router now, you can see, OK, how many tenants are going through a specific router. If you have a single router and a tenant, you can see all the traffic that's going through it. And so now you're starting to get a more granular view of what's going on inside your platform. So this is pretty slick. And then when you kind of go down, you can also see all the routes. Anything that a typical network engineer would want to see, they can see it immediately through a GUI, which is where the entire open stack platform is going. So now what we're talking about is this is the trace request, and this is going to be kind of the most important aspect of the product. What we're seeing here, these are traces that you create. So if you have a user who says, ah, I can't ping between two VMs, what you could do is you create a trace with the dropdown. You pick a tenant or device, whatever you want to do, and then you pick the device name. And you're going to set match conditions. So you say, OK, well, tenant A can't ping between the two VMs, so I want to see ICMP between x value, whatever you're going to put in here. And so when you do that, you'll create a trace, and then you hit the Start Trace button. And it will allow you to see all the flows coming between those specific items that you put in there. So you can see on the bottom it says zero flows traced. When you hit Start Trace, it'll start counting off the flows. And so this is storing them in real time for you. And then what you could do is you click into the actual trace itself. And when you're inside the trace, now you see, OK, well, this was IPv4. It was ICMP. These are all the flows that matched it. You see the specific bi-directional flows, too. Since it's a fully distributed platform, the connections can go anywhere. So you see both return connections. And then if you really want to get some more advanced data, you hit Show Trace Data. And this actually shows the entire simulation of what happened to the packet in that flow. You'll see the topology it went through, the security group rules that it hit, a bunch of things like that that you normally would have to dig around a bunch of log files to pull out. We pull that into a central platform like this. And the next logical step to this is that was real time. Now we're talking about flow history. With flow history, we're storing everything on a cluster in the back end, so a database cluster. And you're able to go back as far as your storage will allow to see exactly how many flows per second. Specific flows that you're looking for, a specific VM. You could do a site-wide search, too, for you could just type a VM name. Or if you even know the UUID, you can type it in. And you can pull all this type of information out. So in this case, we're looking at the ICMP from two VMs, the destination IP. And then you can also show the details. And so we're storing all the details. And this is where it gets really important. When you look at the Flow Visualization Table, that's the actual path the packet took in the network. So this is all virtual stuff, but you still have to simulate this so you can see exactly where that packet hit through every single device and port that's inside the metonet overlay. So here you can see which router, which port, which security group rules it hit inside of the metonet. And then you can see the jump rules here. You see the protocol types. And so this is how you're starting to manage everything that's going on inside of the system. And so this is real operator tools for real clouds. Where all of us are starting to get out of the kind of just playing around with the open stack. And it's very serious now. So we need tools like this. So checking out security chains. It needs to be done in a very fast and efficient way where you don't need a degree in engineering just to see what's going on with your overlay. And that's kind of the point that we're reaching right now. It's still going through the demo here, but you can kind of see this is the return flow. You can see the bridge that it was attached to, what ports in the bridge. And then in this case, it's actually going north-south out of the cloud. So the edge router concept is the open stack concept that we consume. And that shot it out in BGP, and it returned on the different BGP links. So that's the insights platform. I want to show one more thing in the demo here. Also with Flow History, what we're going to give you is metrics, too. And this is super important to everything that you're doing. So when you put in a date range inside of the Flow History, you can see what top tenants, top VMs, if you can identify problem VMs. Some guys saturating your 10 gig links, just doing Tor or Bitcoin mining, whatever you want to do. You can spot that really quickly through this platform. And you can do it of an aggregate amount of time. So you can say, I want to know my whole data for a year. I mean, if you want to store that much data, you can do that with Mem Insights. So the next logical step, though, when we're talking about what operators are doing, that's cool. We're seeing everything that's in the virtual world. But what about the physical world? You want a single pane of glass for tools that you can troubleshoot both the physical and the virtual layer. So we came out with a product and we just released it. It's called Mem Fabric. It's in tech preview right now, but you can definitely take a look at it. What the Fabric module is, it's kind of a plug-in for Insights. And it works in conjunction with open switch software. So the likes of Cumulus, Dell OS10, everything like this, we have an agent that we can run on these Linux switches now. And so what this allows us to do is a lot of different things. But it adds in underlay information to your flow history and flow tracing. So now we start seeing some crazy stuff, and I'll show you that in just a second. Yeah, so the other point to being able to see what's going on with the underlay is the fact that you need to understand if you're having congestion, something that's causing packet drops that you're not aware of that you won't see in the virtual environment. You just see a packet that just went missing. And you're like, okay, well, it was accepted. It, you know, all the rules match, but it still got dropped. So you bring this information in and you can have like a complete picture of exactly what you're doing. So I'll do a little demo of the Mem Fabric. And it's going to look very similar because it's still going to be the Mem Insights interface that you're going to use. But in this case, we have a topology with a spine and leaf switches and a cross topology. Inside this cross topology, we have two tours. They are a bunch of 10 gig NIC ports and one with a 40 gig connection that traverses it. So what we're going to do is we're going to initiate pings between two 10 gig servers. So they're pinging across the switch topology. And then what we're going to do is we start a basically congestion over the 40 gig links. We start saturating the entire system. And so in this case, we have two different pings. They both have two different DSCP values so we can see the different flows inside the system. So when we start the traffic, this actually will slow down VMs because the switch ports are being overloaded. And so the way this correlates into the platform here is, well, you're going to see the high round trip time here in a second. But when you go into the flow history here, now what we're going to add in is the fabric portion of it. So just like we did before, you're going to click through. You're going to see a bunch of different information but once he gets here, he's going to do a match flow. Same thing. So this is how you're searching your database basically right here. So you search it, now we pull back the same information we were just looking at in the last demo, but when you scroll down, you have a show path button now. And so this is where we're going to start getting into the interesting stuff. When you click show path, what it's going to do is it's going to show you the exact path through the entire topology that packet took. And once it clicks through here, I'll show you. And so this is a very easy visual way to see, okay, well, the virtual simulation happened but it never made it. Where did it die? So when you look here, this is the entire topology. All the compute nodes, the spine switches, the leaf switches, everything. You see the blue lines indicate the egress path and then the ingress path is coming back in a different route. But that 40 gig link that we were saturating there, you see it, it's colored red. He's showing you right now that when you click on the actual device, you can see all the ports that are active on it and you can get more information from there. But the key point is you just have a quick visual way. You're like, oh, there's a red box there. Red means bad. So you can take a look. Okay, that's a congestion event that's happening. So now you can, well, there's some future stuff that we'll do that I can't talk about right now, but so with that, you just kind of see exactly what's going on. So red, sorry, red is packet drop events. So in this case, the actual switch port itself is being overloaded. And so it's dropping some packets. The orange means congestion. And so you can, if you think about what you could do with a larger environment, you could start shifting traffic as needed to make sure that you're giving the best quality and service to all of your end users. So he's taking a look at two different flows right now. So we're seeing that was the initial egress flow for the 10 gig side. Now we're looking at the 40 gig side. And you can see, in this case, it's not hitting congestion events because it's going out a different switch port that's not being overloaded. So you can do some cool stuff on your switches there. So this adds into the MEM platform itself and it gives you to what I feel and what all my customers tell me is the kind of tools that they need to, like I said, be able to fully operate the entire OpenStack cloud. So that's going to be it for my demo. It was very quick. I didn't want to put too much fluff into it. Just wanted to get to the details. So please come by, see us at booth CA if you guys want to know more. We can also do a live demo over there. I did videos because the internet was being a little wonky, but so come see me and I appreciate everybody for coming. Thanks.