Recently I've noticed several 10Gb and 40Gb interfaces showing up in the real-time utilization chart with crazy spikes of 46,000% utilization... We aren't seeing crazy numbers like this in the normal statistical charts. It's only in the real-time chart. Has anyone else seen this issue? See the screenshot below for an example:
Here's a different example. This is a VLAN interface that shows a speed of 10Gb and maintains an average receive utilization of 25%, yet the Real-time Chart keeps showing spikes of 2,500%... It's almost like it's multiplying the actual utilization by 1,000 when reporting the data. What gives?
No, my original example was with physical interfaces and the newest example I gave was a VLAN. It seems to be occurring only on high bandwidth links like Port Channels and VLANs where the speed is at 40Gb or higher. And it's not all of them.
The more I look into it, I think it's an issue with UCS Switches specifically, as I'm only seeing it on them. I think it's because the switches have a SNMP bug where they are reporting frequent bursts of incorrect traffic. See the screenshot below and you will see that even though this port channel is 80Gb, NPM is seeing traffic in the hundreds of gigabits... But if we go do a show interface on the device, it only shows traffic of a few hundred megabits.
