Monitoring bandwidth utilization is a balancing act between granularity, router capabilities, and database space. The more frequently you poll, the more load you put on the router, and the more space you consume in Solarwinds' database.
You won't be able to Report with any finer granularity than you poll a target node for. By default, polling is set to 120 seconds. If you need greater granularity, open NPM, go to the router in question, Edit that node in NPM, and set polling to a lower number. Then your reports will be able to pull that data out at a one-minute level, if that's how frequently you poll.
Remember that if you want to go back further in time you'll possibly have to adjust your database compression to happen less frequently. I keep detailed statistics retention data for reporting as far as 7 days back, then hourly at 30 days, then daily at 365. Consult with your DBA folks about data growth & capacity before changing your settings, so you don't overload the servers and use up all your disk space in five minutes.
Seriously. That's possible.
1 of 1 people found this helpful
Also remember that polling every 120 seconds will miss some of your router's highest spikes.
I don't want to overload my router or NPM or database, so I use the Engineer's Toolset and create a bandwidth gauge for several most-important interfaces, then I select the Historical Graph option for each gauge. Finally, I set each one to display that graph in real time on a separate PC's monitors, and configure all to poll every ten seconds.
It's an eye-opener to compare the output of this kind of more real-time polling & graphing to the 120-second polling & graphs in NPM. After comparing the two graphs for some weeks or months you'll get an idea what kind of factor must be applied to NPM's graphs to estimate how much bandwidth is actually being used during spikes that are missed by NPM, which are NOT missed with the Toolset's Bandwidth Gauges.
And last, remember that even the Toolset's Bandwidth Gauges set to poll every 10 seconds are going to miss spikes. So they're not the do-all/be-all/end-all for what's really happening.