01-06-2020 11:09 AM - edited 01-06-2020 11:25 AM
The explanation of the bandwidth command is a little ambiguous to me. "The bandwidth command sets an informational parameter to communicate only the current bandwidth to the higher-level protocols; you cannot adjust the actual bandwidth of an interface using this command." But in practice what I am seeing in monitoring is a 1Gbps configured with the bandwidth command of 500Mbps - peaking at 500Mbps. Is it likely that we're actually getting 1Gbps peaking but the interface is not reporting properly for the bandwidth command? Or is it that somehow TCP is not taking advantage of the full bandwidth because it somehow has been "informed" by the bandwith command? (Never heard of such an informing..) That is could the bandwidth command in IOS XE in any way throttle the available throughput of the interface?
.
01-06-2020 11:36 AM
The bandwidth command does not change the actual bandwidth of an interface. The bandwidth command is used for other things like OSPF to change the cost of an interface. If you want to change the bandwidth, you would need to use the speed command. Also, if the interface is gig and your monitoring device is showing 500 MG, it is most likely that the reporting device is not configured correctly or there is a bug in the software.
HTH
01-06-2020 11:58 AM
Hi @CiscoMedMed,
The interface bandwidth command alone does not perform any throttling of traffic through the interface. It does affect how the bandwidth statistics of that interface are calculated and it is used by routing protocols that would use interface speed as a metric. It essentially overrides the interface speed from a reporting perspective, which can be seen in the statistics of the show interface command.
Throttling of traffic through the interface would be enabled by a separate QoS mechanism such as policing or shaping. 3850 Platform QoS reference here.
A practical use case of the bandwidth statement is for sub rate ISP circuits. As an example, you might have a 1Gbps physical interface hand off from an ISP, however you only purchased half of that speed with a CIR of 500Mbps. The ISP will throttle your traffic with a QoS mechanism (policing) to 500Mbps. By setting the bandwidth statement to 500Mbps on your interface, your intent would be to monitor and report the true available bandwidth according to the limit that is set by the ISP. Note: It is also considered a best practice to use QoS shaping to match the ISP CIR of the circuit, so you the customer would drop the traffic before the ISP does.
If your interface is not also configured with a QoS mechanism then it is possible that your interface is exceeding 500Mbps, If it was peaking at the 1Gbps physical link speed there is likely to be discards on the interface. This would be seen in 'show interface' command output as drops/discards.
Hope this helps!
Thanks,
Rob
05-18-2020 06:23 PM
I had a same problem.
In my case, there is a bug (cisco said it isn't and the bug was classified to 6 Enhancement).
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvm47656
Workaround is not to use bandwidth command.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide