Hi, On Wed, Jan 29, 2014 at 11:25 AM, Marcel Reutegger <[email protected]> wrote: > this works fine for the backup part and partial restore. the only > concern I have is disaster recovery. just replacing the MongoDB > files would be much faster than copying over the nodes from the > tar files to MongoDB.
I would expect the backup recovery speed to be governed by the time it takes to copy over binary values, as they take up by far the most space in large repositories. Thus the performance of an Oak-level recovery should be at least within the same order of magnitude as a lower-level file copy. Alternatively, it should be fairly straightforward to adapt the existing Oak backup code to work also with MongoMK target, as most of that code uses only on standard NodeState functionality. That way we could maintain a separate backup MongoMK repository, and simply start it up for a full recovery without the need for any copying. BR, Jukka Zitting
