Consider this scenario: Everything is up to date on firmware version, down to the blade adapter. Additionally the recommended/supported enic/fnic drivers are fully updated. The cluster is running close to the max number of hosts (28 for example out of 32). Also, assume the VMware Tools version is updated to the correct version.
There is no ping latency to the blade server or to the mgmt vmkernel port. However, the users are experiencing latency and performance issues (example 4000ms). In other words, the data path to the VM Hosts is experiencing latency.
At this point head to VMware to troubleshoot further: