Unsolved
This post is more than 5 years old
9 Posts
0
1643
VMware Linked Clone performance issues?
We have a large service provider hosting thousands of VMs with Netapp VAAI FASTclones and NFS. We tried to convince them to use VMware Linked Clones as Best Practice, but they justified not to use them with the following arguments:
“VMware linked clones are fast clones without VAAI support:
- misaligned writes (creates performance inefficiencies)
- read performance suffers especially for longer chain lengths
Just with the misaligned writes, the service provider's VMware admins worry about massive performance issues and better do not use it. “
Is it true, that we(VMware) are suffering on misaligned writes when using VMware Linked Clones?
Do we(VMware) have a read performance issue especially on longer Linked Clone chain lengths?
Any advice is welcome...
Andreas
itzikr
28 Posts
0
October 10th, 2014 01:00
In the VERY old days (esx3) , if you created VMFS partitions via cli, they were unaligned by default, its not an issue since 2009 ☺
sullybags
45 Posts
0
October 10th, 2014 01:00
Andreas
what they are referring to is VMFS partition alignment.
page 5
http://www.vmware.com/pdf/esx3_partition_align.pdf
to the best of my knowledge this is not a issue on EMC storage due to the block sizes we use.
its been a few years since i checked this, but the document above should be a good starting point.
itzikr
28 Posts
0
October 10th, 2014 02:00
Well, VMware recommend to run a refresh of the pool whenever the linked clone reach around 13% of size of the parental image.
Apart from it, check your array, I assume you are using fast cache.
Ando2
9 Posts
0
October 10th, 2014 02:00
Thanks, are you guys aware of any read performance issues on longer VMware linked clone chain lenghts?