Paul Hoffman wrote:
At 1:50 PM -0700 9/15/05, Michael Thomas wrote:
Which is pretty much the elephant in the room, I'd say. How
much of the net traffic these days is, essentially, not in
any way standardized, and in fact probably considers ietf
old and in the way?
Not sure why this is an elephant; who cares?
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?
I have seen numbers that
show that a huge percentage of traffic is P2P of various flavors, but I
haven't seen anyone saying that this is having any negative effects.
I don't think this is _entirely_ true: p2p stuff definitely
has, um, interesting effects on, say, voip at home, and some
of the p2p apps -- especially the earlier ones if I recall
correctly -- had some pretty nasty effects on various networks.
Are we to believe that they are largely self-healing problems
as bad p2p apps will eventually correct themselves since it's
in their interest? Is it reasonable to believe that there is
enough general clue that they could be expected to do that?
And the collective clue of the ietf is not really needed to
help this along?
I'll note that many protocols -- good and bad -- spring from
somebody's head. Some of them become successful too. Very
successful. And ietf has no say about them at all. Is this
the new reality?
But for layer 7 protocols, file sharing
may be the only major market that has wholly ignored the IETF.
This isn't that unusual really, but what facinates me
is that the reverse seems true as well.
Yes, if one that has bad congestion control becomes popular. But, given
the mindshare of BitTorrent these past few years, that seems pretty
unlikely.
But surely BitTorrent isn't the last one that will come
along. I guess the base question is this: is the net robust
enough to really allow experimentation with flash crowds of
millions of alpha testers? So far it has, but 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? That is, is
there a price for this benign neglect that has yet to surface?
Mike
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf