[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Performance of CVS version SVN
From: |
Peter Toft |
Subject: |
Re: Performance of CVS version SVN |
Date: |
Thu, 26 Oct 2006 07:00:21 +0200 (CEST) |
On Thu, 26 Oct 2006, Michael Haggerty wrote:
> Peter Toft wrote:
> > I just made benchmarks of CVS 1.12.12 versus svn 1.1.3 (r12730)
> > for "update to tag/date/current" operations for a 2500 files project with
> > 6 years timespan in CVS. Approx 62 MB.
> >
> > I found that CVS always outperforms Subversion typically with 2-3x
> >
> > Can anyone acknowledge that performance? Any similar experiments.
> >
> > It must be noted that the SVN archive was made with cvs2svn - using the
> > CVS project.
> >
> > I will publish my results when I have a bit more time...
>
> This is not terribly surprising; svn is not especially fast at
> checkout/update operations. At the recent SVN summit there was a lot of
> talk about how to improve the situation by using a different repository
> storage format. Part, though probably not most, of the extra cost is
> writing the second copy of the pristine text in the working copy.
>
> But, when you publish the results, I would like to suggest the following:
>
> - Indicate which SVN filesystem you used (bdb vs. fsfs). They differ
> considerably in performance.
I have fsfs - bad or good?
That came from cvs2svn
> - Indicate what access method you used (file://, svn://, svn+ssh://,
> http://, https://). They differ considerably in performance.
I have to use svn+ssh://, but I have also make localhost measurements
using file://
> - To be fair, include a timing comparison of "cvs tag" vs "svn tag" :-)
Good idea - any other operations also?
> Michael
>
Best
Peter Toft, Ph.D. address@hidden http://petertoft.dk