Hi,

so far we have four possible places we could move to. I would particularly
like to thank Solomon and Bas for offering to host this privately!
However, I have some concerns that these sites could be subject to the
same kinds of occasional bandwidth problems that seemed to trouble Alfred
university four weeks ago. It's true that those can never be ruled out,
but sites that are explicitly dedicated to providing CVS services should-
at least in theory- be less prone to fall for these reasons, simply
because they are less subject to seasonal changes.


First, here is a summary of the sites we could be hosted at, in no
particular order:
(1) square.a-eskwadraat.nl, hosted by Bas on a 100 Mbps connection.
(2) shaft.shaftnet.org, hosted by Pizza
(3) sourceforge.net, hosted by VA Linux
(4) berlios.de, hosted by GMD FOKUS
(5) Stay at cs.alfred.edu, hosted by Chris Lansdown


And here's how my connection to these sites is on this particular time of
day:
Site    Av.Ping average speed/home      average speed/work
(1)     120 ms  24.5 KBps               290 KBps
(2)     310 ms  7.5 KBps                10 KBps
(3)     210 ms  60 KBps                 60 KBps
(4)     65 ms   85 KBps                 290 KBps
(5)     175 ms  9 KBps                  17 KBps

Of course, for CVS, other things like local I/O performance
(writing/reading from HD, making diffs etc.) is also relevant, as would be
regular backups. As is the continuing availability of the service, which,
as has been pointed out, might be questionable for (3).

Clearly, from my point of view BerliOS or Bas' site would be ideal.
Note that the 290 KBps values are, as far as I can tell, without caching-
BerliOS is run by a different division of the company I (occasionally)
work for, after all, but the good connection to (1) came as a pleasant
surprise.

I'd recommend that those of you who use (or want to use) CVS regularly or
occasionally should try to assemble a table like this one of their own. My
favourite candidates might well be completely inadequate from your
locations, and we should try to avoid unneccessary susprises like these...


llap,
 Christoph

--
There are lies, damn lies, and statistics. -- Winston Churchill


Reply via email to