Hi Everyone,
some time ago we evaluated the snapshot features from Cloudstack for the combination Ceph + KVM + Cloudstack and found some space for improvements. Our goal is to provide snapshot backup for our customers which is easy to handle over the Cloudstack API and provides incremental/differential features. We recognised that Cloudstack uses qemu-convert to extract the data from the raw device. This means that every time a backup gets triggered 100 % of the used data gets transferred to the secondary storage and not just the delta. For big volumes (500GB to TB volumes) which are highly used, the backup process can last very long and uses much space on the secondary side. >From our side it would be nice to have a differential and/or incremental >backup feature for our Ceph volumes. So we took a look around and found this as a possible solution: http://www.backy2.com/ This is a python tool which can do differential backups from Ceph snapshots and store them on a filesystem or in an S3 Storage. Sadly there is no REST API for this tool which makes the management of it a little bit inconvenient. We were wondering if this tool, some parts of its functionality or the main approach of it could be integrated into Cloudstack to improve the snapshot functionality. I checked the Github issues and did not find any similar tickets for that. So maybe here are more people who needs this functionality or have already some solutions for this. BR Alexander
