Hello.
On Thu 2002-08-22 at 11:16:53 -0500, Ben Collins-Sussman wrote:
> Michael Price <mprice@atl.lmco.com> writes:
> > Blair Zajac writes:
> > > I do need to sweat the log stuff doing the incremental backups.
> > >
> > > I plan on replacing the db backup solution with the svnadmin dump based
> > > backup solution due to the large disk space that the db based backups
> > > are using, I indicated my the first message in this thread.
> >
> > I guess I'm already lost then. With several of my repos at home, the
> > gzip'd dump is several times LARGER than the gzip'd repository itself.
> >
> > Where exactly is the space savings? (or do I not understand what your
> > doing?)
>
> Yeah, I was going to ask this same question! A dumpfile doesn't have
> *any* compression in it at all, the way our repository does. If a
> file changes in some revision, the repository stores it as a diff, but
> in the dumpfile the fulltext is always written out.
>
> What's going on, Blair? I don't understand the itch you're
> scratching.
I assume the idea is that you can make an incremental backup of the
repository via dump, but not via binary copy.
In other words: Yes, a dump will be larger than the repos, but when
you want to keep a backup every day, the binary copies for a month
will takes 30 times the space, while the incremental dumps will be far
less.
Well, that presumes that you cannot make (reasonably) an incremental
copy of the binary, what I do not know.
Regards,
Benjamin.
- application/pgp-signature attachment: stored
Received on Thu Aug 22 18:31:22 2002