A challenge with many network management systems that collect and display events and statistics is they leave the processing and analysis of that data completely to the operator. For something known with certainly to be a problem, systems usually generate an alarm. However if a network has sub-optimal performance, the issue might be subtle and the administrator would have to co-relate a number of things to root-cause the issue. This is both time-consuming for the network administrator and potentially error prone.
Would'nt it be great if the system could highlight any anomalies automatically? In the Autopilot 'Insights->Pulse' feature we attempt to solve this by having the system analyse the statistics that it collects, and flag things that are potentially a problem. Its a one-page 'pulse' of the network to detect its overall health.
The following screenshot shows the current set of scenarios the system looks for:
we try to detect and highlight things that could lead to performance issues, and are either due to mis-configuration (an AP with no SSID mapped), power issues (low uptime of APs), switch-port config/cabling problems (Ethernet port auto-neg problems) or operational issues (APs with firmware not matching the one on the master-AP).
The advantage of this sort of analysis is when the administrator logs onto a network where say clients have been reporting issues, this becomes a one-stop-shop for a number of potential symptoms at one go. It saves time from having to switch through various dashboards and statistics pages, digging through logs etc. Time saved in root-causing leads to a faster fix of the problem, and a better end user experience.
Please share other things you would like the system to detect automatically.