RE: Performance of CVS version SVN

[Peter Toft]

On Thu, 26 Oct 2006, Arthur Barrett wrote:

> Michael/Peter,
> 
> > - Indicate which SVN filesystem you used 
> > (bdb vs. fsfs).  
> 
> Things like the Server OS and client OS and network transport (ie:
> T1000) would also be useful.
> 
> However I personally think that benchmarks of SCM systems are not very
> helpful.  

The reason for studying it is that I have a large CVS project, which 
requires SSH access over the net to a remote server where it takes 10 
minutes to do all the "cvs update" operations. People are nagging about 
that, i.e. I needed to see of e.g. Subversion is faster or not.
 
> I've never seen an analysis of SCM implementation cost / benefit that
> indicated that the performance of a system affected the implementation
> cost or the benefit to the organisation or people using it.  In fact
> mostly I see "usability", "accessibility", "ensure integrity", "track
> relationship between changes", "manage change", "determine metrics (by
> reports)" etc as being the key drivers.
> 
> If you can include something in your benchmark about how you envisage
> the numbers may help or relate to cost/benefit analysis or some way of
> interpreting them - I think that would be very helpful.

See above :)

> Finally you mentioned that your stats on svn were based on a cvs2svn
> script - it could therefore be assumed that you know more about / work
> more with / CVS than SVN.  If you are publishing the benchmark on a
> non-partisan site (ie: not cvs.org, cvsnt.org etc) making that sort of
> stuff clear helps transparency.  If you are publishing to a partisan
> site then it'll simply be assumed that the info is biased that way
> anyway.

Point taken,
I know CVS quite well, but I have no "darling". Subversion has some good 
properties (by design) so it is obvious to analyse that.

Also thanx to Michael Haggerty - I can comment later on his postning.

Best


Peter Toft, Ph.D. address@hidden http://petertoft.dk