compdigit44
asked on
ESXi 6 TPS
I know by default TPS is disabled for inter-vm sharing for security reasons. Our DMZ servers resides in a separate VM cluster from our production VM. My question is has anyone enabled this for inter-vm sharing and if so how much memory saving did it by you?
ASKER
Does the changes require a reboot?
It does not require a reboot, but lots of options here to confuse you... because since it was disabled, they are no various different options based on Host version, and salting!
see here
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2097593
see here
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2097593
ASKER
IN the following article what does the term memory sharing between VMs with the same salt mean?
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2097593
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2097593
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
and this was due to a very low risk security concern, that VMware decided to turn if OFF by default.
Many of our Clients have re-enabled this now, because it does help, especially with oversubscription e.g. it does save memory.
After the patch we have seen 200GB in shared pages, which have resulted in memory contention issues. (on some workloads, so the savings are different per host, cluster and environment)
It is believed to be low risk see here
https://kb.vmware.com/selfservice/search.do?cmd=displayKC&docType=kc&docTypeID=DT_KB_1_1&externalId=2080735
Here is a tool you can use, to check if you still have enough memory, if it's disabled versus enabled!
http://www.running-system.com/still-enough-host-memory-inter-vm-tps-disabled/
So in response to your question.
1. Yes we have enabled it. (The business considers it low risk versus memory saved!)
2. It saved 200GB. (memory saved on this workload).
Otherwise you've just got to increase your budget and put more memory in hosts!