[e2e] common congestion controller for TCP connections

Joe Touch touch at ISI.EDU
Tue Oct 29 10:42:52 PST 2002

To add to the discussion,

IMO, the primary difference between CM and TCB-sharing approaches is 
what happens with a single packet.

in CM, a single congestion control system processes every packet

in TCBI, conventional per-connection congestion control processes each 

In both cases, information can be shared among a set of sequential or 
concurrent connections, and in both cases the aggregation and timescale 
(per packet, at connection start, etc.) of sharing is determined by 
policy, not the architecture. CM allows non-TCP applications to benefit 
from TCP-'like' congestion control; TCBI is simpler to use from the 

E.g., E-TCP is TCB-sharing where a policy of "connections with the same 
src/dst should be aggregated as if a single connection". There are, as 
you point out, other policies. TCBI includes a suggestion that is 
similar to that of SPAND, that some congestion information can be shared 
among endpoints on a LAN, FYI (see "Implications").


More information about the end2end-interest mailing list