We recently expanded our VMware Infrastructure on two Dell
PE 2950 with 16 GB each. ESX 3.5.0 (build 82663) was installed on both and we also utilized
a SAN environment. We started installing
VM’s with one Windows 2003 Server. The first server also had our VC installed on it,
and has 2048 MB installed (Shares: Normal, Reservation: 1024 MB, Limit: 2048
MB). The second server is going to be a DC for our domain and has the identical
What we noticed was with the installation of the second W2K3
server, alarms began to appear regarding memory usage. It occurred when the
second server was booting up and then the alarm would clear itself. In doing
some research on VTMN, we noticed that this was not unusual for VM’s running
W2K3, as per the following post: http://communities.vmware.com/message/501136#501136
What is causing some consternation is the fact that we never
saw any alarms when we were running in a single host environment running ESX
3.0 that had half the memory– and there were 3 W2K3 VM’s running on that
platform. Also, we never saw the alarms
prior to adding the 2nd Windows VM on the newly installed ESX host.
Based on the post listed above, this is to be expected with VM’s running
Windows – but the fact that we never saw these alarms before and that there
doesn’t seem to be a reason for over-commitment on the memory is giving us
We decided that we wanted an "official" explanation from VMware so we called support and opened a service request. The technician assigned stated (after checking with other engineers) that the way ESX 3.5 handles guest memory will produce a 'spike' at boot up that may result in alarm - even though over-committment is not occurring.
In nutshell, ignore the guest memory usage alarm that occurs at boot up on Windows VM's.