> But there is no /backup/ technology to do that now, that I know of. A

I don't know why I have to keep repeating this.  Crashplan does backup this
way, but you can't restore a sparse file sparsely from a crashplan backup.
I have a ticket open with their support.  They say they'll address it, but I
don't count on such things before I see a result.

Is crashplan the only backup system in the world to calculate byte
differentials?  I doubt it.  But you seem to think so.


> checksum on the whole file won't tell you what /block/ changed.  One
> would need checksums on /each/ block.  I don't know of any backup
> system
> that does that. 

That's what I'm asking for.  Anyone who knows any tool that can do something
like this.


> The backup log would be a significant fraction of the
> filesystem, or if sparse, a significant fraction of the data. Lets say
> you have a 1k block and you want to monitor for changes, and use a
> 160byte mac to ensure no collisions on changes having the same sum. See
> the problem?  

No.  I don't see the problem.  You say 160byte, but I was thinking more like
this:
Store a checksum for every 1Mbyte, and one more for the file as a whole.
This way, you run the risk of checksum collisions on the 1M blocks (a very
small percentage risk), and the overall checksum is a cross-check, which
allows you to detect if such a collision occurs.  If a collision occurs,
there's no choice about it, you have to resend the whole image, but that
would happen very rarely.

And the checksums are less than 0.1% of the total backup size.


> Not to mention the issue of computing the checksums
> during
> backup, looking them up in a database, which has its own overhead.  The
> backup system could become the major load on the server.

More than backing up the whole file?  I think not.  This "overhead" is a
time and effort saver, not loser.


> Of course, the versioning filesystem doesn't do it that way.  It just
> keeps pointers to a copy-on-write set of blocks that have changed,
> rather just like virtual memory, starting with the root inode (actually
> plural).  One only needs to compare the two root inodes to find what
> blocks have changed between them. At some gross over-simplification,
> just 'Lather rinse repeat' for the rest of the inodes in the
> filesystem.
> You get the point.

If ZFS or something else that does Copy On Write is available in Windows,
and enables sending byte differential backups, please let me know.

_______________________________________________
bblisa mailing list
[email protected]
http://www.bblisa.org/mailman/listinfo/bblisa

Reply via email to