Article written by

Sébastien is a Cloud Engineer at eNovance. His daily activities are mainly focused on OpenStack and Ceph. Quite often, he rotates between Devs, Ops and Pre-sales teams. During his free time, he loves playing Magic the Gathering cards!

5 Responses

  1. Simon Leinen
    Simon Leinen at | | Reply

    Thanks a lot for the article! It’s super useful for us to understand the limitations of this combination, which we use (mostly happy with our decision :-).

    Flatten volumes created from snapshots [...] Just keep in mind that
    flattening clones increases the space used in your cluster.

    Yes, indeed – especially when a user spawns a large number of VMs – say 500 – from a given snapshot. With flattening, you’d need 500 complete copies. Without flattening, only one “base” volume is needed, but each VM can still manipulate their volume in isolation, thanks to RBD’s copy-on-write feature. We have (scientific) uses case for this, so I guess we’ll leave rbd_flatten_volume_from_snapshot set to false – which will hopefully remain the default.

  2. OpenStack Community Weekly Newsletter (Nov 22 – 29) » The OpenStack Blog

    […] Back from the summit: Ceph/OpenStack integration […]

  3. mganguli
    mganguli at | | Reply

    Hi. It is great to see that there is a solution to the inefficient booting solution to test. What branch can we download to get that?

    thanks.

Please comment with your real name using good manners.

Leave a Reply