===> .ne. <===: Asymmetric CU-SeeMe Performance

Joseph M. Izen (joe@utdallas.edu)
Thu, 19 Sep 1996 08:29:48 -0600


I have had several CU-SeeMe sessions where one direction works fine, but
the dropout rate in the other direction is unusable. Even using similar
audio codecs! In almost every case, I was able to verify that the Internet
routing of the signal was different for each direction. Why should this
be? You'll probably need to ask an Internet routing expert. I've
encountered this when aiming at a National Lab site in the US. When I
transmit from a generic location, the Internet routers realize this signal
is bound for a national lab, and pick up ESnet (Energy Sciences Net) which
has much better performance than the public Internet. Unfortunately,
signals originating at a lab very quickly are identified as headed for a
generic site and are hence unworthy of being carried by ESnet and get
shunted onto public carriers. When I have done netcasts of scientific
meetings for the labs, I have circumvented the routing problem with the
help fo ESnet by using CU-SeeMe reflectors along the ESnet backbone across
the US and good jumping off points to Europe. Then local reflectors at
universities aim for the nearest ESnet reflector, and finally clients
attach the the local reflector on their own campus or a nearby (net-wise)
campus.

-Joe Izen