From: Scott Schremmer (scotths@cs.washington.edu)
Date: Wed Oct 20 2004 - 02:59:16 PDT
This paper attempts to deal with the evolving types of networks by
designing a new protocol that works well over networks with high bandwidth
and latency. This is an important issue to consider as this
characteristic is common among satellite and optical links.
An important part of this new protocol is header information
providing details of congestion that a packet encounters. This provides
much more timely and accurate information on the state of the network than
the older method on relying on dropped packets. The protocol also
separates a congestion manager from a fairness manager in the routers.
The congestion manager deals with congestion issues and the fairness
manager provides the bandwidth in a fair manner. (Possibly taking into
account bandwidth assigned to various users and parts of the system).
The protocol is shown through analysis and simulation to function
quite well. They claim it works as well or better than TCP in all
situations, including situations in which traditional TCP works fine. It
would be interesting to see how well this would work in a large network.
This archive was generated by hypermail 2.1.6 : Wed Oct 20 2004 - 02:59:16 PDT