For about 2wks our 'snmp/ping notification' software has alerted me to issues with my Exchange 2K7 U3 server, which is a VM running on ESX 4.0U2. The notification software is IPCheck, in case any of you all are familiar with it. We configure "ping" and "snmp" sensors. Ping sensors are just that..to report on the VM being down (or unresponsive for a prolonged period and unable to respond to pings). The snmp sensors monitor disk usage so we can be alerted when data volumes are running low.
Anyway, around 8-10pm, when backups run against Exchg (Avamar B/U), CPU runs high. According to vSphere Performance, the VM only runs about 70% utilization, but within the guest, it's max'd out (100%). I have 8GB RAM and 2 vCPUs config'd for this VM and, to be honest, only having around 220 users, with minimal mailbox storage (limits ARE imposed), those specs seems a bit high to me.
Sometimes IPCheck messes up and the sensors need re-done. (not sure why that is...just an old pgm actually) I did re-do the sensors, but still am getting notifications. I do also notice OWA somewhat unresponsive during the 8-10pm timeframe as well. And even during the day my Outlook 2010 client is sporadically unresponsive.
So, my question is this - how can I best config my Exchg VM so CPU doesn't run so high. My assumption is that it is semi-unresponsive during this B/U window and not really "down". If I can get my VM config'd so CPU doesn't run so high I'm thinking that would mitigate this, not sure. On the other hand, I don't really want to config this VM with a bunch of resources it doesn't need 85% of the time, even though I do have plenty of resources to give it. Any suggestions are appreciated. FYI - vSphere/VM performance monitoring/testing is not a strong suit of mine :)