On 16-sep-2005, at 1:00, Michael Thomas wrote:
I'm not sure; maybe it's really a mutual non-admiration
society, and everybody's happy? But it's an elephant
insofar as it's pretty darn big trafficwise, and the
fact that ietf doesn't seem concerned?
Why should the IETF be concerned about traffic?
Now of course that doesn't mean our amateur protocol designers get it
right every time... For instance, when you become a Gnutella
"ultrapeer" you'll very likely create some nasty congestion in the
outgoing direction of your ADSL or cable connection because when you
receive one request, the program wakes up and sends out copies of
that request over several dozen TCP sessions at the same time, which
will invariably overload the buffers at some point between the host
and the DSL/cable connection.
we're layering
more and more stuff onto the net too -- like voip -- that
are pretty sensitive to average expectations (I'm thinking
about things like Vonage, not managed services). Is that
a danger for the overall internet architecture?
Anyone who thinks he or she is going to meet real time constraints
over a random path across the internet that spans multiple ASes has
more problems than the IETF can possibly fix.
And no, I don't think it's reasonable to restrict p2p bulk data
transfer to accommodate voice over the public internet.
(Maybe at some point people from Apple would like to share the
results of them setting a DSCP code point / type of service in the IP
header of voice packets.)
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf