On Thu, Dec 13, 2018 at 3:24 PM Suchitra Herwadkar <Suchitra.Herwadkar@veritas.com> wrote:

Hi Nir & RedHat team

 

  1. Cloned VM support in RHV.

There are two types of clones as per my understanding

  1. Full Cloned VM: Independent copy of a virtual machine that shares nothing with the parent template after the cloning operation.
  2. Dependent Cloned VM: It is a copy of a virtual machine that shares virtual disks with the parent template. This type of clone is made from a snapshot of the parent which is qcow2 image. All files available on the parent at the moment of the snapshot continue to remain available to the linked clone. Ongoing changes to the virtual disk of the parent do not affect the linked clone, and changes to the disk of the linked clone do not affect the parent.

 

We want to know what the customer preference ( 1 or 2) as both have pros & cons. What is the guidance given by RHV around this?  How do we recommend customers to patch the template disks when multiple VMs are cloned from it?  Kindly provide more details.


Customer can use both types of clones, based on their needs. If having the dependency on the 
a shared template is fine, and the user is ok with having qcow2 format and at least one snapshot,
and prefer faster cloning, they can go with this option.

If the user wants to use raw format which can give better performance, and ok with waiting for 
a slower clone, copying all the data to new image, they can go with full clone.

So I think on your side you should support both options, so users can make the best choice for their
needs.

 

  1. Is there a way to we can mount a QCOW2 image using QCOW2 mounter and have a block level view of the data ?  We read about libguestfs but this is filesystem view.

In 4.3 we will use NBD protocol to allow block level view of an image, including all snapshots.

You will be able to do full or incremental backup, or restore raw data to qcow2 image.
See:
 

  1. We would like to know when we can expect “Allocation Regions API” . I’m referring to

https://bugzilla.redhat.com/show_bug.cgi?id=1622946

This is part of incremental backup effort:

This is marked for 4.3 at this point in time.  I presume we will get it as part of 4.2.X release as well.

No, this requires major changes in engine, vdsm and imageio, which will be available only in 4.3.

The new API is documented here:

We would like to get early feedback, before we implement it.

Nir