Steve Tempest
asked on
ESXi 4 outperforming Esxi 5
Hi,
I'm a little stumped. I've to two hosts one is vsphere 5 enterprise plus and the other is vsphere 4 advanced.
I use these hosts to run a citrix workload. The other day I noticed that the execution time of a cpu intensive task on the version 5 host was taking 2 times longer to run that on the verison 4 hosts.
After some testing I can see that the application is hitting a single core and that the amount of time to process the same transaction between servers is 2:20 on v4 and 4:20 on v5.
I've ruled out the build of the citrix server by migrating one from v4 to v5 and the longer result occures.
So the only differences I can see is that the citrix servers have hardware version 7 on the v4 host and version 8 on the v5 server.
I'm 99% sure it's a cpu thing, it makes no difference if there are other users on the system during testing the results are the same as with no users on the system.
V4 physical
v5 physical
The application is running some basic type scripts which is looping and processing some data in memory and I notice that the memory for the process increases by about 100mb during the processing... it's just that the v4 server does this cpu intensive task much faster than v5...
And yes I have checked everything the network/config of guests the lot.
I'm stumped?
I'm a little stumped. I've to two hosts one is vsphere 5 enterprise plus and the other is vsphere 4 advanced.
I use these hosts to run a citrix workload. The other day I noticed that the execution time of a cpu intensive task on the version 5 host was taking 2 times longer to run that on the verison 4 hosts.
After some testing I can see that the application is hitting a single core and that the amount of time to process the same transaction between servers is 2:20 on v4 and 4:20 on v5.
I've ruled out the build of the citrix server by migrating one from v4 to v5 and the longer result occures.
So the only differences I can see is that the citrix servers have hardware version 7 on the v4 host and version 8 on the v5 server.
I'm 99% sure it's a cpu thing, it makes no difference if there are other users on the system during testing the results are the same as with no users on the system.
V4 physical
v5 physical
The application is running some basic type scripts which is looping and processing some data in memory and I notice that the memory for the process increases by about 100mb during the processing... it's just that the v4 server does this cpu intensive task much faster than v5...
And yes I have checked everything the network/config of guests the lot.
I'm stumped?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
I think this is one, for VMware Support!
ASKER
yeah I'm thinking so... there isnt a known issue with this that I can find. whats funny is if I run a cpu benchmark utility on both v4 and v5 from within a guest they benchmark close to the same.
We have basic support ... I'm not sure how helpful they will be.
We have basic support ... I'm not sure how helpful they will be.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks for the holp
ASKER
The esxi v5 hosts were clean builds so the vmfs filesystem should be fine.
None of the guests have reservations on any host. I have added reservations and it hasnt improved the speeds.
Cheers