SIOCv1 vs SIOCv2 - VM share aggregation/pooling

Hi All,

Hopefully someone can fill me on this one.

At the moment, I have ESXi 6.5 clusters (upgraded from 6.0) that have SIOCv1 enabled.  Ultimately, I want to be able to control shares at a VM level, not VMDK.

Because a VM’s total SIOC shares in SIOCv1 are pooled (and can be ‘borrowed’) across all its VMDK’s, I calculate the per VMDK shares for a VM by dividing the desired amount of total VM shares amongst it’s VMDK’s.  For example, if I have a linux VM with 10 VMDK’s and I want a total of 1000 shares, I give the VMDK’s 100 shares each.  This means that I can give an equal percentage of the datastore’s shares to each VM, regardless of the number of VMDK’s they have.

I want to simplify this though and move to SIOCv2 policies.

What I haven’t been able to find out though, is whether or not the shares allocated though a SIOCv2 policy are pooled for a VM in the same way as they are for SIOC v1.

I can see when applying the policy there is a new object “VM Home” – but I have no idea what this object is controlling. If I attach a VM storage policy to a vm that gives 1000 shares, it configures 1000 shares per VMDK, so need to make sure that these shares are not able to exceed a VM maximum limit of 1000, nor are they pooled to give a VM total of x * 1000 shares.

If not, then I’d be best staying on SIOCv1 and applying these calculated SIOC shares using orchestrated powershell scripts.

Tags (1)
0 Kudos
0 Replies