I’ve gotten a lot of questions from customers about how to monitor their NSX-T edges in AVS. Since the edges are considered management components customer don’t get access to the edge VMs in AVS (or most other clouds). This is done to protect the customer from breaking their environment by accidentally doing something to the VMs. One could argue that read access to the VM would make gathering statistics much easier for the customer but that’s just not how the system is designed. Luckily for customers they do have read access to the Edges nodes, the T0s and T1s through NSX. This access lets us do interesting things. With some API calls it’s possible to pull all the interface and CPU stats for the edge VMs and T0 uplinks. There are plenty of other things to pull as well but these are what most customer ask for. Through some python and a couple of Azure services all of these can be collect, stored, and graphed on a dashboard.
The repo is here: https://github.com/khensler/nsx-t-stat-kusto. The dashboard is included as well. This is not an enterprise monitoring solution but will collect data and display it for you. The script will run wait 10 seconds then run again. In my environment this means a data collection once every 17 seconds or so. If you plan on using this you may want to adjust the polling interval to something more reasonable. The readme in the repo has instructions on how to use it.