ESX CPU at very low average but if I drop core count on SQL server it strgules
On ESXi 5.5 I have 2 x hosts, each host has 12 Cores, 2 socket x 6 core.
We have a bunch of servers running on them and have a total of 70 CPUs allocated out.
Our main DB server currently has 8CPU allocated, however if I drop it down to 4 is struggles.
Given that, over the course of the year and also month by month, we don't use more than an average of 10%
of the total CPU processing power on the ESX hosts, should this be happening?
I'm trying to figure out how many cores we need in our new ESX hosts and I'm confused by the fact that
although we're not using a lot of the overall CPU power, the SQL server struggles if I reduce the core count.
Should it not just take up more of the unused cores?
We have a bunch of servers running on them and have a total of 70 CPUs allocated out.
Our main DB server currently has 8CPU allocated, however if I drop it down to 4 is struggles.
Given that, over the course of the year and also month by month, we don't use more than an average of 10%
of the total CPU processing power on the ESX hosts, should this be happening?
I'm trying to figure out how many cores we need in our new ESX hosts and I'm confused by the fact that
although we're not using a lot of the overall CPU power, the SQL server struggles if I reduce the core count.
Should it not just take up more of the unused cores?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Do you mean RAM? There is a lot of RAM on the server, the SQL server uses all we allocate as you would expect. What I dont understand is this. Say on the host on which the SQL server is running has a total of 12 available Cores, we've allocated 8 to the SQL server and and the hosts stats say that that the total CPU usage, over an average of say a month is 18%. The we drop the CPU count on the SQL server to 4, why dont we see the host CPU avg shoot up as it consumes all of 4 CPUs in line with seeing the SQL servers CPU graph shoot up?
So basically, when I look at the graph on the individual machine I see it getting wacked, but the overall host performance doesnt seem to mimic it as much as I think it should.