We have a newly deployed UCS domain, and we have added it to Solarwinds Orion to keep an eye on bandwidth usage. I'm finding that the portchannels and ethernet connections between the blade servers and the IO Modules is quite high with a lot of spiking. Most of the time some of the PCs/ ethernet connections are showing almost total saturation of their 10 Gbps connections. What is strange is that there is then very little traffic being reported out of the virtual ethernet and HBA connections for the blades, and coming out of the IO modules to the FIs. The blade servers seem to be operating normally, but I'm unclear on whether this is causing a performance issue. Here's the equipment involved:
6332-16UP FIs
1 40-Gbps uplink from each IOM to the FI
2304 IO Modules
B200M5 blades with VIC 1340 adapters
4.0.4b firmware
RHEL 7.6 with two vNICs and fabric failover enabled.
Has anyone seen this before, and have any clue what is going on here?
