[e2e] Comparing different TCP flavours?

[Jon Crowcroft]

panos gevros' thesis looks at ensembles of different TCP's 

TCP sweet (add i. m d)
TCP sour (mimd)
TCP bitter (gigd)
TCP stout...

i don't know anyone who has looked at the real problem (lots of flows
in the various different startup, and post congestion event
strategies) - most papers appear to be about the rather rare and
ironically terms "steady state" :-)

having said that, recent papers (e.g. OSDI one this week) point out
that the average download object size is increasing due to p2p...

jon