Let's get more data, please (Was: Let's remove mips, mipsel, s390, ... )

[Dirk Eddelbuettel <edd@debian.org>]

Adam Heath <doogie <at> debian.org> writes:
> On Tue, 22 Feb 2005, Dirk Eddelbuettel wrote:
> 
> >         files.downloaded  percent
> > i386             1285422  70.5079
> > all               504789  27.6886
> > powerpc            17754   0.9738
> > ia64               10111   0.5546
> > sparc               3336   0.1830
> > arm                  850   0.0466
> > alpha                507   0.0278
> > hppa                 204   0.0112
> > mipsel                91   0.0050
> > m68k                  15   0.0008
> > mips                   7   0.0004
> > s390                   4   0.0002
> > total            1823090 100.0000
> 
> These numbers show a cross-section of users who use this particular mirror.
> It is not represenative of the world as a whole.  Far from it.

Thank you. You are about the fifth or sixth person who took the time to 
explain that to me. But guess what? I knew that too --- but these were *the 
only numbers* I had ever seem.  As I just said in another follow-up, the 
debate would be helped greatly if we had data from more primary mirrors 
and/or popular hosts, and longer time frames.  

We actually need that data to have the qualitative discussion to determine 
at what point the cross-product of nArches * mSourcePackages is too much for 
our infrastructure.  

Lacking access to the actual apt-get usage data, I just posted another
complimentary analysis using the data based on popcon.debian.org reports. You
may find it interesting.  It does not contradict the above in any meaningful 
way  (esp. once you acknowledge that it is based on a smaller sample, and a
potentially biased population of core Debian developers and supporters aware
of popcon, as opposed to mom&pop just using apt).  

Regards, Dirk