Thick Provisioned Template Deployment Why?

Hi,

Weve finished trialing extremeIO a few weeks ago, we have a large VMware environment and template deployment speed was very impressive with the ExtremeIO.

Traditionally all of our templates are thin provisioned but when we switched them to thick eager zeroed we saw a massive speed boost in their deployment. (4 x faster). We are also seeing this on our VMAX now we have tried it all be it to a lesser extent.

Could anyone explain the technical reasoning behind the speed boost we see when deploying VMs from thick eager zeroed templates over thin provisioned templates? Are different VAAI primitives in use or is there a different underlying cause im missing?

Thanks

Jay

Responses(1)

itzikr

28 Posts

0

February 12th, 2016 15:00

Hi Jay,

first, i'm glad that you like the array.

eager zeroed thick VMs will be faster to clone from because VMware has less metadata to process Vs thin, make sure you also set the XCOPY chunk size to 0256kb as per our best practices

Optimizing XCOPY Block Size for EMC XtremIO | Itzikr's Blog

now, having said that, if you are running vSphere 6 and your VMs OS's are windows server 2012 / 2012 R2 / Windows 8 and above, IF they will be configured as thin VMs, you can also benefit from a semi automated space reclamation inside the guest OS's

https://itzikr.wordpress.com/2015/04/28/vsphere-6-guest-unmap-support-or-the-case-for-sparse-se-part-2/

lastly, try to take the VM that is used as your temaplte, convert it to thin vmdk and then convert it to template again and deploy a VM from it, let us know the speed of the cloning Vs cloning from an eager zeroed thick VM.

thanks

Itzik Reich

CTO - XtremIO

View All

No Events found!