What I find often is that high latency alerts for datastores are tied into PLANNED moves. Big chunks of data moving to meet a planned need. Are your datastores showing high latency without cause? for extended periods of time?
I have the internal vcenter team provide guidelines on what they consider latency for read ps and write ps and make those HARD threshold KPIs in vcops. Then when an alert happens its a matter of understanding planned vs unplanned, then its a matter of drilling into data with your experts.
IMHO - Being admin on vcops doesnt mean you know the environment better than your teams... it means you have useful data to share and then develop the best levels together and the SOP for when latency occurs and the steps to follow. We are working thru stuff like this now... and testing the Symmetrix adapter. Currently that isnt addressing the HBA level so we miss some key data points, but we are working on the txt adapter and other methods to get better data in vcops.
