cmatchett
asked on
hot add cpu / memory vmware exchange 2010
Is anybody aware of any issues that exist when hot adding cpu and memory on a virtual machine running server 2008 r2 / exchange 2010?
nope in fact I believe I have done it. Are you experiencing issues? If your vmware tools are up to date and the hot add has been enabled in the virtual machine settings you should be good to go.
ASKER
need to upgrade the vcpus on this particular server. Hot add memory and cpu isn't currently enabled.
i would like to enable it when i am turning the vm off to add the vcpus - for future changes
i would like to enable it when i am turning the vm off to add the vcpus - for future changes
OK. The VM will need to be shut down to enable it. But then it should work moving forward.
ASKER
what size of deployment did you do this on?
one server typical install. roughly 500 mailboxes. I made sure I had a good backup before doing so. I recommend that for you of course as well.
ASKER
ok, thanks. It might be a week or two before i get this done. i will report back
Personally, on a production machine, I would not Hot Add Memory or CPU!
ASKER
Any particular reason? My outlook is that technology has moved on now with vmware etc and applications have now been written to better cope with the facility to hot add?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Ok, i have come across a DAG that consists of 2 servers and 4 databases. 2 passive copies on each server, so thats +20% already on the processor.
these two servers only have 4 vcpus. there are 2000 users.
CPU is nearly constantly sitting above 90 % on each server, so when one of the server fails the server hosting all copies of the databases doesn't do to well.
i understand that utilization should be around 40% on a dag member
these two servers only have 4 vcpus. there are 2000 users.
CPU is nearly constantly sitting above 90 % on each server, so when one of the server fails the server hosting all copies of the databases doesn't do to well.
i understand that utilization should be around 40% on a dag member
Add more servers.
ASKER
The only options available to me is to increase the vcpu / memory
That may not resolve the 90% cpu issue, adding CPUs, is not linear....
ASKER
I suppose the process will have to be that i first upgrade the vCPUs.
I was looking at this pdf
http://www.vmware.com/files/pdf/exchange-2010-on-vmware-best-practices-guide.pdf
I was going to up the vCPU from 4 to 8, although 12 might be the correct amount.
I should really use the exchange 2010 calculator but arranging to run the profiling tool could prove difficult
I was looking at this pdf
http://www.vmware.com/files/pdf/exchange-2010-on-vmware-best-practices-guide.pdf
I was going to up the vCPU from 4 to 8, although 12 might be the correct amount.
I should really use the exchange 2010 calculator but arranging to run the profiling tool could prove difficult
Just be cautious, and check metrics that adding too many vCPUs may not have the desired effects and make performance worse.
Many VMware Admins, add too many vCPU and oversubscribe VMs, making performance worse, because they can!
Many VMware Admins, add too many vCPU and oversubscribe VMs, making performance worse, because they can!
ASKER
doesn't the profile analyzer add load to the exchange servers whilst it is running?
Yes it does.
ASKER
I will perform a switchover of the active databases when upgrading each server?
As opposed to running the DAG maintenance scripts?
As opposed to running the DAG maintenance scripts?
ASKER
when you add a vcpu to a vm, do you have to..
turn it off
add the vcpu
turn it on
turn it off
turn it on again?
Like change block tracking?
turn it off
add the vcpu
turn it on
turn it off
turn it on again?
Like change block tracking?
You will generally find, after adding a vCPU it will be added automatically via Plug and Play, BUT, you then may find, a message which states, New Hardware Found, Restart Yes or No.
That message does not go away, and is displayed to any admin which logs in, so best to restart again, to avoid any restart during production hours!
That message does not go away, and is displayed to any admin which logs in, so best to restart again, to avoid any restart during production hours!