Sub-interface has same Tx/Rx load as physical interface (different bandwidths)

[Larry Sullivan]

Hi,

My sub-interface (200 Mbps BW statement) is inheriting the Tx and Rx load from my physical interface (300 Mbps statement).  Is this some sort of bug?  Is there a workaround?  Anyone come across this?

GigabitEthernet0/0/0 is up, line protocol is up
  Hardware is BUILT-IN-2T+6X1GE, address is 0
  Description:xxxx
  MTU 1500 bytes, BW 300000 Kbit/sec, DLY 10 usec,
     reliability 255/255, txload 192/255, rxload 180/255

GigabitEthernet0/0/0.643 is up, line protocol is up
  Hardware is BUILT-IN-2T+6X1GE, address is
  Description:xxxx
  Internet address is xxxx
  MTU 1500 bytes, BW 200000 Kbit/sec, DLY 10 usec,
     reliability 255/255, txload 192/255, rxload 180/255

[Larry Sullivan]

No workaround for this huh?  Anyone else notice this behavior?

[Richard Burts · Accepted solution]

I had not noticed this but when I think about it I find it pretty logical. If you think about it you have a physical GigEther interface that is operating as a trunk interface. The vlan sub interfaces are logical/virtual interfaces which are differentiated by a vlan tag number. But all tx and rx activity is done by the physical interface. Your sub interface is not doing its own tx or rx and so there is not a good way to report tx or rx at the sub interface level.

HTH

Rick

[Larry Sullivan]

Thanks for the response.  Let me clarify my use situation.  We have an NNI interface with a provider with different services/VLANs.  Total NNI is 300 Mbps of which additional service within that is 200 Mbps (capped by provider).  So it would be nice to see load for each service within the NNI.  But I see what you are saying.  Makes logical sense, although I don't see it as being too complicated of a feature for Cisco to implement.  I would have expected it to be implemented at this stage in networking evolution as things have been getting more and more complicated for years and of course will continue to do so.  But I guess not many people have noticed this or considered it enough of an issue to request a feature request on.  I just wanted to see if I was missing something, or if there was some sort of trick or workaround.  Thanks again.    

[Richard Burts]

I am glad that my response was helpful. Thank you for marking this question as solved. This will help other participants in the community to identify discussions which have helpful information.

I understand your use case. As long as you are looking at txload and rxload on the interfaces I believe that you are stuck. But I wonder if in various network management solutions there is not reporting software that would break down the load in ways that would be helpful. I do not have a specific example but believe that there are some out there.

HTH

Rick

[Larry Sullivan]

Solarwinds reports the correct BW usage although we do have another issue with that that may or may not be related.  (Tx on one end reports 50 Mbps more than Rx on the other end).  But yes, SNMP should report correctly.

[Richard Burts]

It is good to have confirmation that Solar Winds does report usage figures for the sub interface. Is the reported usage close to the limits for the vlan/subinterface? Is it possible that the provider is dropping some traffic? Is it consistently 50 Mbps difference and always in the same direction? One of the challenges in comparing sent on one end to receive on the other end is to make sure that you are really measuring exactly the same time interval.

HTH

Rick

[Larry Sullivan]

Good point about the polling interval.  I checked Solarwinds and they are all the same intervals though.  Usage is running at about 4/5th of the circuit cap/sub-interface BW.  It is always in the same direction.  It is not a consistent 50 Mbps, it varies according to traffic.  More traffic, higher phantom traffic amount.  If provider is dropping traffic, it isn't service impacting that we are aware of.  50Mb of non-service impacting traffic doesn't seem likely.  This is multicast, so I'm looking at that angle as well.  

[Richard Burts]

I wonder a bit about your statement that the circuit is running about 4/5. Given the fact that most of our data traffic is bursty if the average use is 4/5 then it suggests that there might be small intervals where the traffic is at or beyond full capacity.

HTH

Rick