Open for Voting
over 2 years ago

Create a New Historical Bandwidth Utilization Report

It would be nice if with in Solarwinds you could run a report that would allow you to select your Internet connection interfaces and run a report that shows utilization during business hours.  The report would allow you to select the interfaces to monitor and also select the time frames from which you want to see the data.  The data would also be represented in both bps and % utilization.  In addition, this report would have to keep the data separate in someway so that the data would be compressed over time.  The issue we have is getting an accurate picture of our bandwidth usage to pass on to management.  We have to create a custom report and even then data gets compressed as time goes by so by the end of the month, even though you select only working hours and exclude weekends the data is still distorted because of the data compression after 4 hours and a day and a week and so on.

  • Its really a good option to have this.

  • I've requested similar ideas in the past, and this one's a good one.  Throw in the ability to look back through a data warehouse for Internet or specific WAN port utilization across five or ten years, while they're at it!

    Then work on the compression and filtering; I'm only interested in accurate data, and specifically interested in peak utilization.

    When I run the Engineer's ToolSet Bandwidth Gauges pointing at a WAN or Internet interface, set to poll every ten seconds, the output is MUCH different than the data presented in NPM's 120-second polling of those same interfaces.  I'd prefer the data were identical--at least as far as peak utilization goes.

    Then when looking up historical data, even the last three months, the peaks seems to easily be lost in the compression.

    What the Bandwidth Gauges show in 10-second polling may be 30% - 50% higher than the data shown in NPM's 120-second polling, and that's not going to generate confidence in NPM's "true" bandwidth reports.

  • I use a custom property on the interface for this purpose.  It's manual when you first create the node but it helps keep my process refined. We have threshold alerts setup for these interfaces.  I have a scheduled report for each interface too.  It's quite cumbersome and having something canned or automated would be nice.

  • I would like to second the comment regarding accuracy over time.  This is a big issue for lots of network management systems and if addressed I think could be used to further differntiate Solarwinds products from the competition.  The issue is summarising or averaging statitics over time.  The longer the time frame, the lower the reported utilisation, so suddenly, the 500k of peak traffic in my voice queue over the last hour has turned into 200k over the previous day etc, however it can go both ways....Our Internet circuits do practically nothing overnight, so this lowers the overall utilisation when looking back over >24hours, but on our internal WAN links we have overnight backup and replication which maxes out the circuits, so skews the utilisation the other way!  As mentioned, this is an issue when I want to size circuits by looking back in time and we end up in the "producing lots of customised reports" scenario that dcal has described above. In these situations, a min/avg/max chart for Netflow and CbQoS charts would be very useful, just as is done for interface utilisation in the NPM.  We also use Fluke's Netflow Tracker and that has a time filter that allows you to specify the times of day you want to include - it's quite useful, but personally, I'd prefer the min/avg/max, especially for cbQoS charts of individual queues.

  • This is a great request, well defined. It's these types of use cases that really help us as we look at extending the reporting functionality in NTA. Would love to see more like this from the community.