Re: Size of hotbackup.py files versus svnadmin dump files

kfogel@collab.net wrote:

> The dumper didn't exist when we wrote hotbackup.py. I don't know how
> much more expensive it would be, in terms of execution time, to do a
> dump than to just copy the files.

I haven't tried to benchmark the difference, but it does look like
hotbackup.py + tar is much more cpu and i/o intensive. Whenever I run
hotbackup.py, my drives are churning away more intensely and for longer
than just a straight dump.

tarring up the directory structure created by hotbackup.py is clearly
more cpu and i/o intense too. Presumably because tar has to recurse down
the tree with hotbackup.py, rather than just operating off a single dump
file.

>>This would save an enormous amount of space. E.g., a tarred/gzipped
>>dump of my repository is just 28Mb versus 429Mb from hotbackup.py!!!
>
>
> Well, why aren't you gzipping the hotbackup results too, if you want
> to do that comparison? :-)

I just did... rather interesting results:

svnadmin dump --> 129Mb file
hotbackup.py --> 429Mb directory tree

svnadmin dump + tar --> 88Mb file
hotbackup.py + tar --> 28Mb file

Big, big differences, especially considering that out-of-the-box (on
Debian at least), hotbackup is set to save 64 previous backups. So the
difference in disk consumption is potentially 64 x the difference!

...Edmund.

...Edmund.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri May 16 21:17:04 2003