Dean-
Well, it depends by what you mean 'TCP limitations.' If by that you mean
something like stop-and-wait behavior when more than one packet is lost in a
window, that is fixed in modern TCP implementations. If you mean TCP's slow
start behavior at the beginning of a connection and after a timeout, that is
actually a feature rather than a limitation in that it keeps the Internet
stable during periods of congestion. If you mean TCP's reduction of cwnd
following a detected packet loss, this is also considered a feature that
provides network stability.
Having said this, there are cases where TCP performance over certain types
of links leaves something to be desired. The IETF PILC working group is
chartered to make recommendations and consider well understood and less well
understood mechanisms for improving performance over lossy, asymmetric, and
low rate channels (long delay having been pretty thoroughly examined in
tcpsat). If promising mechanisms are identified that require protocol
modification or development, the door is open to re-charter PILC to do that
work.
I encourage you to read the PILC drafts and mail archives and participate in
the work if you see something that needs doing.
--aaron
> -----Original Message-----
> From: Dean Martin [mailto:[email protected]]
> Sent: Thursday, July 22, 1999 1:30 PM
> To: [email protected]
> Subject: tcp enhancement
>
>
>
> I think it should be interesting to investigate efforts on a protocol
> that is enable to cope with TCP limitations. up to now this type of
> protocol has been proprietary. TCP enhancements brings new features to
> TCP, but still now the implementations will be dependent on the
> Operating system Vendor. Whereas a normalized protocol x that converts
> the TCP into another x packet will be quite interesting in future
> applications . the question remain open?
> _________________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.com
>
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:55 EST