Re: [patch] delayed acks decrease performances in congestioned networks

David S. Miller (davem@dm.cobaltmicro.com)
Sat, 14 Nov 1998 17:27:24 -0800


Date: Sat, 14 Nov 1998 17:31:30 +0100 (CET)
From: Andrea Arcangeli <andrea@e-mind.com>

I don' t think that setting the delayed ack timeout to 0.5sec is a
good idea. The max timeout should be calculated in a smoothed way
in function of previous received packet.

The 0.5sec is the upper bound, which is specified by the RFC's.
If you look in tcp_input.c:tcp_delack_estimator() precisely what you
are asking for is there.

There are some other problems with fast retransmit which I have fixed
very recently and should take care of a lot of problems some people
are seeing on lossy links. I shall merge these to Linus before
Monday. So let's test those fixes out in 2.1.129 or so before
thinking about providing a way to disable delayed acks...

TCP's retransmission strategies, especially when SACK is available on
the connection, should even make the most broken of networks perform
decently.

Later,
David S. Miller
davem@dm.cobaltmicro.com

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/