Forum Discussion

BenBross's avatar
8 years ago

Bandwidth reporting cumlative over time with multiple interfaces

 

LM Dev's,

I would like to be able to run a report over multiple interfaces representing total bandwidth used cumulatively for example.

I have 13 offices with 1 mpls  connection at each office.  I would like a report that would show me the total per day mpls traffic for the 13 offices as one line in a chart/graph over the course of say 3 months.  To determine if I can see company wide trends in utilization.

please let me know if you have any further questions.

Thanks,

Ben

  • 3 minutes ago, RaheelandBenPowerCoupleOfTheYear2017 said:

    I AGREE THIS WOULD BE AWESOME LETS MAKE IT HAPPEN ALRIGHT GUYS?

    qft

  • Anonymous's avatar
    Anonymous

    FYI there is a new interface statistics datasource in the works. There are literally millions of interfaces monitored by LogicMonitor so making changes to this DS involves A LOT of testing and reviewing to make sure it works and doesn't break things. It's coming. I've gotten a sneak peek at it and it's incredible. Getting total bytes in/out per poll cycle is on my list of things it needs to do. Having that datapoint should allow a "sum" consolidation in reports (as well as elsewhere) that would allow for a report that would yield the sum of all bytes in/out of an interface over a larger timeframe.

  • On 4/6/2020 at 1:21 PM, Stuart Weenig said:

    FYI there is a new interface statistics datasource in the works. There are literally millions of interfaces monitored by LogicMonitor so making changes to this DS involves A LOT of testing and reviewing to make sure it works and doesn't break things. It's coming. I've gotten a sneak peek at it and it's incredible. Getting total bytes in/out per poll cycle is on my list of things it needs to do. Having that datapoint should allow a "sum" consolidation in reports (as well as elsewhere) that would allow for a report that would yield the sum of all bytes in/out of an interface over a larger timeframe.

     

    Hi Stuart, do you know if the ability to see total bytes in/out per poll cycle was ever added to any interface datasource. Looking at the snmp_network_interfaces datasource, it doesn't seem to be the case. I too have a customer who is interested in seeing the total amount of data over an interface over a certain time period.

  • Anonymous's avatar
    Anonymous

    Yeah, that's what the inOctets and outOctets datapoints are.

  • Anonymous's avatar
    Anonymous

    If you want it over a period of time, you'll need to build a report and use the sum option. That will sum all bytes in and all bytes out per poll cycle over the timeframe.

  • I realize the original request in this thread is from 5 years ago, but since it popped up again, perhaps still useful to respond :).

    Since this is a known set of devices, the problem originally stated can be solved with a widget that has each interface as an element (hidden optionally), then a CDP with the sum of all interfaces as the displayed element. Admittedly, the widget builder should be a smarter about selecting datapoints via query so you could automate this task based on interface attributes, but you could still create the widget and make a point to update it when the network expands.  In theory, the smart-widget maintenance bit could be done via the API.  If you need it emailed, setup a dashboard report to handle that piece (this may not have been possible 5 years ago).

    If you are literally looking for total GB/day for one or more interfaces, then that is not going to happen with a widget or alerts (both of which should be possible -- I am getting increasingly spoiled by how flexible Grafana is :)). That could be accomplished by building the widget, then use the API to download the widget's data and transform as needed. 

    That transformation could also be done within a Groovy datapoint using your Nth copy of the same API code (maybe the same, or maybe you edited 10 of them to fix a bug and the other 5 still have the bug -- what common programming feature could possibly help with that?).

  • 3 hours ago, Stuart Weenig said:

    If you want it over a period of time, you'll need to build a report and use the sum option. That will sum all bytes in and all bytes out per poll cycle over the timeframe.

    Thanks, I think I was getting confused about the derive metric type returning a rate. I was thinking it might be necessary to multiply the result of the derive datapoint "InOctets" times the polling interval of the datasource. If the In/OutOctets datapoints are just returning the amount since the last poll, then summing the result of all polls will give me what I need.

  • Anonymous's avatar
    Anonymous

    You can multiply by ##POLLINTERVAL## to get it back from rate.