Application Awareness Overview

In this demo, we'll show how the Juniper Data Center Assurance platform helps identify network issues when there's a service impact in the data center.

In our demo environment, we have a two-spine, two-leaf topology with multiple hosts connected to the leaf switches. These include both bare metal servers and ESXI hosts. Virtual machines running on the ESXI hosts are displayed at the top layer. For this demo, we'll simulate traffic between the two Ubuntu VMs, a server and a client. And the server will stream video using VLC. The client will access that stream. Then we'll introduce network congestion to see how Juniper Data Center Assurance detects and reports on the issue.

Here is our Ubuntu server VM where we have the VLC streaming server. The client VM is accessing that stream over port 8080 and we see that that stream is playing smoothly on the client.

In the App Aware page within Juniper Data Center Assurance, we can view traffic flows across the data center. Services are listed and organized by throughput. We can drill down and locate our video service running on port 8080. And when we click on it, we see the service topology showing the video service running the Ubuntu server VM, clients accessing the stream from the Ubuntu client VM, and the full network path through the spines and leaves.

And at this point, the data center health is normal and no anomalies are reported. So, at the moment our video is running smoothly and everything is fine. To simulate congestion, we switch to a test VM and start running high bandwidth iperf traffic between a test client and a server.

Before congestion, we see the video in the client playing smoothly. Once the test traffic begins, we observe the video slowing down and becoming glitchy.

This mirrors a real world situation where users may complain of video buffering or slow application performance. The root cause could be the server, the service, or as in this case, the network. Once we stop the test traffic, the video quickly returns to normal.

A few moments later, back in the Dashboard view, we see that an anomaly has been reported. A hot interface alert on one of the leaf switches. This was caused by the congestion. Clicking on this anomaly takes us to the Impact Analysis screen. There we can see the specific leaf affected, a list of services impacted by the anomaly. In our case, it shows two services - port 8080, our video stream, and the iperf traffic we generated caused the congestion. Apstra also lists the clients affected during that time frame, which was from 12:00 to 12:15 p.m.

If no one noticed the anomaly in real time, but a user later reports issues, for example, glitchy video around 12:00 p.m., we can still investigate. Using the Impact Analysis feature, we select a wider time range, say 11:15 a.m. to 12:45 p.m., and scroll through to the reported window.

Even though the anomaly is cleared, Apstra retains historical data. So, we can go back to 12:00 to 12:15 p.m., and we see the service impact, and we confirm that an anomaly was present and the video service on port 8080 was affected.

This demo illustrates how Juniper Data Center Assurance can help identify and trace network-related service impacts like video jitter caused by congestion. Even if the issue is temporary, the platform's impact analysis and anomaly history allow admins to go back, correlate user complaints with actual events, and resolve the root cause efficiently.