Hi,
I have created three resource pools as follows:
Production
High
expandable reservation
40000Mhz limit
High
expandable reservation
25600MB limit
Dev
Normal
Expandable Reservation
40000Mhz limit
normal
expandable reservation
25600MB
Build/Test
Low
expandable reservation
16000mhz limit
low
expandable reservation
14336mb limit
Should I now limit and reserve per VM or let the pools deal with the resources?
I have currently left the vm's as default.
Some of our VM's are heavy hitting on CPU memory so should I limit them per VM?
Thats kinda a personal call with your environment....
It depends on the service levels you need to ensure to all environments. If the heavy hitters are in production they can affect the dev and build/test environments. If you set limits on the heavy hitters, it might impact thier performance.
Depends which is more important to you.
Another thing to remember is this only matters in the event of host contention. If the host has enough resources to serve everything up - everything runs at full pop.
Hi,
thanks for your reply, I was wondering more what most people did in terms of leaving VM's as default resource allocation and let pools manage them or limit vms indiviually in the pools?
regards,
Paul
All the virtual machines within a resource pool will be entitled to equal shares of the pool's processor and memory. If you are happy with that then don't change the virtual machine shares.
Within a resource pool, if you want different virtual machines to have different entitlements to processor and memory, then go ahead and change the shares of each virtual machine. This is most easily done by clicking on the resource pool, clicking the Resource Allocation tab, and changing the settings in the Shares column.
I personally let my resource pools handle it... We have 300+ VMs though and I dont want to have to keep track of manually edited ones and like to keep things standard.
In my shop if a couple VMs are really using up a large large amount of resources we usually put them to physical hardware.
Thank you for your reply.