Monitoring a Cisco ACI-Based Infrastructure


Dave Hegenbarth introduces the Cisco partnership alliance and discusses the benefits around monitoring application-centric infrastructure with SevOne.


Hi, I'm Dave Hegenbarth, Director of Systems Engineering, for Global Strategic Partnerships at SevOne. Thanks for joining. The days of service providers and large enterprises buying rack after rack of specialized network computing storage, to launch a new service or application are coming to an end. Our customers and the market as a whole, are on a journey to software defined everything, where network, compute and storage, along with automation, and orchestration, will allow them to innovate and deliver applications and services much quicker through a virtualized infrastructure.

At SevOne, we're committed to making that journey with our customers and our technology partners, and provide complete performance visibility, through the technology transition that is about to happen. In this white board session, I'm going to introduce to you our alliance partnership with Cisco Systems, the application-centric infrastructure and the benefits around monitoring such a deployment with SevOne. Let's go to the white board.

On the whiteboard I've drawn out at typical SDN/APIC infrastructure. APIC is the application policy control that we see at the top here. The job of APIC is actually to control the flow of services and traffic through the underlying 9K infrastructure. This is the Cisco 9K switches that provide high speed east, west paths of traffic throughout the network. With this 9K infrastructure provisioned, and with the controllers controlling it, we have the ability to deliver applications very rapidly throughout the enterprise.

I'm going to draw an example of this. My example starts with a marketing guy who needs a new website. That marketing guy is thinking about his website and the fact that he needs to make that website a reality very quickly. There's also specific users within the infrastructure, or within the environment, or within the organization that will be able to use this, while others will not. Part of building this out is the control of flow traffic through certain devices in the network. There are certain application devices that we're going to need to instanciate to make this all work.

I might need a firewall for security, so I'm going to have my firewall virtualized, probably. I'm also probably going to have some sort of virtualized IPS, and lastly, I'll have maybe, a load balancer to handle a lot of traffic, if the website gets very popular. On the back end of this, we have the services that are going to serve up the content for my marketing website. I'm going to draw a couple databases as well. These are the resources that we are going to need to provide that end user application or marketing website.

My user will go in what we call an end point group. All of these objects will be part of that end point group or a policy that gets formed and pushed from the controllers down to the 9K network, such that the traffic traverses through my firewall, IPS, and load balancer, as it should, and delivers the services with maybe, the quality of service enabled, that delivers this service in a rapid manner to the end users who need the service.

SevOne comes in, in the middle here, as a performance monitoring solution. Now part of the performance monitoring is going to through the APIC controllers themselves. There's a wealth of knowledge about how the network is performing and a wealth of metrics being collected about it as well, and being reported via the APIC controller.

There's also some other things left over that we probably want to capture. Those are some of the stats from our virtual devices that we've put up in the network, some of the stats, maybe, from the servers themselves that are putting it up. Also, we might want to take information from other things, such as an edge gateway.

We've left the virtual 9K arena to hit an edge network, probably out to the Internet, or possibly over to Cloud providers. We see a lot of these web resources and our customers being divided between a private Cloud and a public Cloud like Azure or AWS. SevOne needs to go and be able to grab information from all of these things. How do we do that?

First of all, we have written to the northbound APIs and the APIC controller. Why did we do this? We did this because we wanted to be able to grab real time inventory, what's in the network. We also wanted to grab real time topology, because this could change as more rules are added to the controller. So we wrote APIs, so that we would have inventory and we also would have topology.

With both of those already in SevOne, we have the ability to go back and then use traditional monitoring technologies, such as SNMP, to actually pull both logical and and physical devices. We're also going to ask for the flows from this 9K infrastructure, NetFlow, to allow us to understand who's talking to who, for how much, what types of protocols, and so forth.

Also, all of this is being supported on top of a Cisco UCS server, running a hypervisor, where these VMs will live and provide these virtual services. This too need to be monitored and SevOne has the ability to poll the UCS server. We also have the ability to collect logs from the UCS server, so we have a good understanding of how this environment is working. We're also traditionally polling these edge gateways, SNMP, probably taking NetFlow back in. Lastly, we actually have APIs that allow us to monitor performance of Cloud providers, like AWS.

All of this put together allows us to build a dashboard with graphs and charts in a single pane of glass. It might have up here a graph, and I'm just drawing this roughly for my firewall. I have another graph that might be from my IPS. I would have graphs for my load balancer, such as connections per second. I might also have graphs that represent the overall health of the 9K infrastructure. I would get those directly from the APIC controller.

One of the great things about SevOne is we bring all of this information into a single dashboard. We create baselines that are understanding and normal for every KPI, whether it's connections per second, number of people hitting a web server, etc., and we're able to bring that into a single pane of glass dashboard, and, or alert, and send you notices when something is no longer normal. All of this brought together in a single dashboard will add tremendous value in understanding the performance of the application that I just fired up. Thank you for taking the time to watch this white board on the integration of SevOne and the Cisco-centric infrastructure. Thanks.