[plt-scheme] Threading vs. System Threads

I haven't been able to post much lately, but I did see the thread lamenting the
lack of native threads under MzScheme.  I originally viewed this as a big
disadvantage to the MzScheme implementation, but lately I've been seeing
evidence that this may not be the case.

Consider the performance of the Yaws server (Erlang's equivalent to the
PLT Scheme server) in comparison to Apache version 2.0 (the new
"threaded" version).   Yaws is able to handle thousands more parrellel
sessions than even the much-vaunted Apache.
http://www.sics.se/~joe/apachevsyaws.html

The conjecture is that the use of OS-level threads for concurrency leads
to lower performance than Erlang's internal concurrency.  I like to imagine
that MzScheme may share some of these positive features (Matthew,
please feel free to correct me! :-)

The intelligensi amongst the PLTers probably already know all of this from
attending MIT's LL2, but we mortals can always enjoy the slides and
presentation materials (e.g., http://www.sics.se/~joe/talks/ll2_2002.pdf).

Regards,

-Brent

P.S.  Vive la PLT Server!