[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Size comparison with CVS

From: Ben Collins-Sussman <sussman_at_collab.net>
Date: 2003-10-07 17:35:27 CEST

Lele Gaifax <lele@nautilus.homeip.net> writes:

> CVS repos: 4.0Mb
> SVN repos: 7.2Mb
>
> Digging inside the /db, I notice that the strings file, the one I'm
> assuming contains the actual sources, is 3.9Mb.
>
> db $ ls -lug
> totale 7198
> -rw-rw-r-- 1 lele 241664 5 ott 20:12 changes
> -rw-rw-r-- 1 lele 8192 5 ott 20:12 copies
> -rw-rw-r-- 1 lele 8192 5 ott 20:12 __db.001
> -rw-rw-r-- 1 lele 270336 5 ott 20:12 __db.002
> -rw-rw-r-- 1 lele 327680 5 ott 20:12 __db.003
> -rw-rw-r-- 1 lele 737280 5 ott 20:12 __db.004
> -rw-rw-r-- 1 lele 16384 5 ott 20:12 __db.005
> -rw-rw-r-- 1 lele 1282 5 ott 20:12 DB_CONFIG
> -rw-rw-r-- 1 lele 687998 5 ott 20:05 log.0000000086
> -rw-rw-r-- 1 lele 499712 5 ott 20:12 nodes
> -rw-rw-r-- 1 lele 565248 5 ott 20:12 representations
> -rw-rw-r-- 1 lele 24576 5 ott 20:12 revisions
> -rw-rw-r-- 1 lele 3985408 5 ott 20:12 strings
> -rw-rw-r-- 1 lele 262144 5 ott 20:12 transactions
> -rw-rw-r-- 1 lele 8192 5 ott 20:12 uuids
>
> Is there any explanation of this difference?

Yes. It's a transactional database. You should expect some overhead.

The strings table is about 4 megs, as expected. That's the main
"meat" of your original file data. The other tables take up about 1.5
more megabytes; that's all the information about directory structures.
(CVS never needs to store information about paths). The __db files
take up about 1.3 more megabytes; this is just shared memory segments.
They're essentially 'scratch' work areas that berkeleydb uses to aid
concurrent accesses. And the one logfile in use is about .7 megs;
this is the journal that allows you to run 'svnadmin recover'.

So in this case, your file data is still about 4 megs, and the
database overhead is another 3.2 megs.

On the other hand, I imagine that if you converted 400 meg CVS
repository to SVN, the SVN repository would *not* have an additional
320 megs of database overhead. I imagine that the ratio of overhead
to total repository size approaches zero as the total repository size
increases.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Oct 7 17:37:28 2003

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.