Reporting on Network Device Resources Being Monitored

shocko

I have been auditing my network devices to see:

1. Which devices are not being monitored at all
2. When they were last refresh for resources on the device to monitor
3. Which resources are being monitored

For item 3 I see a lot of inconsistency namely that one some devises the L2 topology is monitored, on others not. On some the L3 topology is monitored, on others not etc. For example, here's one of my critical routers:

This device should have it's routing table monitored. I'd like to create a report on all network devices and all resources being monitored. Better still if I could report on the resources that are available to be monitored but aren't.  

Of course, I will investigate the DB myself to see if I can find an answer. My feeling is I will be able to report on what's being monitored as they will be in the DB but what's missing is retrieved via SNMP. If that's the case I will need some custom code or a custom poller. 

rschroeder

Your request/suggestion about seeing what's available to monitor and reporting on whether it IS or is NOT being monitored is a good one.  I'll be watching this thread to see what others suggest.

In the mean time, I've found it simple to bypass any inconsistencies by going to the page to manage nodes, then sorting by Machine Type in the left column.  Once there, I select all the Layer 3 devices that are actually performing Layer 3 services (meaning that although a switch CAN do L3 services, you may not have set it up as a router, so there's no need to include it in this particular example), and then choosing the poller/polling that should be applied against everything in that group.  Select L2 and L3 and anything else you want, then apply it.  NPM will check the snmp community strings to ensure it has access to the data, and then you'll have a uniform polling solution for all your L3 devices.

Then I select all the Machine Types that don't have L3 services enabled (or that are not even capable of doing L3 services), and choosing the appropriate L2-only monitoring for their polling.

Yes, it's kind of a low-level, brute-force way of getting the job done, but it really gets the job done in bulk.

I do this for every machine type, and then I work with the teams setting up new nodes and teach them how to monitor each one correctly.  There's nothing like automation, though, to ensure everything is monitored correctly without manually doing it, without relying on human error to make things incorrect.