Hi Sage,
Would qcow2 let you store the incremental portion of a backup without the
actual image? A common use case would be:
- take snapshot1 on cluster A
- copy image to cluster B, possibly without writing to an intermediate
file.. e.g. 'rbd export ... - | ssh otherhost rbd import - ...'
and later,
- take snapshot2 on cluster A
- stream snapshot1->2 incremental to cluster B
Well, qcow2 is designed for storing image files on disk, not as a
"streamed" file format - so it is not covered by the specification.
However, I don't see what should keep us from using it the way you describe.
The only drawback is that standard qcow2 tools will not be able to work
with the incremental file on its own - they would need the base image to
be file based.
Basically a qcow2 "incremental" file contains:
- string holding the path and filename of the backing file
- lookup table that tells which blocks are included in the file
- actual data in 512 byte blocks
In this "streaming" case we would not have a path and file to the
backing file. Instead we could store something like
rbd:pool/imagename@snapname instead of a traditional path and file name.
--
Jens Kristian Søgaard, Mermaid Consulting ApS,
jens@xxxxxxxxxxxxxxxxxxxx,
http://www.mermaidconsulting.com/
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
