I'm running VMware Server 2.0.2 on my dual core Athlon server with 4 GB RAM and a RAID1 with two 400 GB SATA hard disks. This server's running three VMs at a time.
The host system is a Debian 5 x64 with the latest kernel and all updates installed. Besides VMware Server it doesn't run anything else.
The VMs do use non-fixed hard disk images. I'm running two VMs with 768 MB of RAM each, the third one uses 1.5 GB of RAM, so there should be another GB of free RAM for the host system.
Two VMs have a Ubuntu 9.10 x64 installed, the other one uses Debian 5 x64.
My problem is the very poor performance. In one of the VMs I'm running Apache with mod_rails (Phusion Passenger). None of the VMs do have to handle very heavy load. So after a time of idle the Passenger goes to sleep. Waking it up again takes up to 45 (!) seconds during which the VM doesn't really respond anymore due to the load generated while waking it up again. The load meter in the VM peaks up to a 10.00, which, in my opinion, can't be normal. On a (non-virtualized) test system I can't see such a behavior, so it has to be the VMware Server, doesn't it?
Sometimes even a simple SSH connect to one of the VMs generates a very high load, up to 8.00.
Someone told me that it is possible to direct a precise amount of CPU power and other resources to the VMs but I really don't know what to look for. Unfortunately Google didn't tell my either.
Any help is appreciated.