[e2e] timeout with TCP/SACK
lxu2 at unity.ncsu.edu
Thu Aug 21 18:53:50 PDT 2003
Experiments in paper "Modeling TCP Reno Performance" by J. Padhye et. al.
(Trans. on Networking, Apr. 2000) show that "time-outs constitute the
majority or a significant fraction of the total number of loss indications".
But those experiments use TCP/Reno. Are there any experiments using
TCP/SACK? With TCP/SACK, are timeouts still the major part of all loss
More information about the end2end-interest