Re: revision RFC 2001 (draft-ietf-tcpimpl-cong-control-00.txt)

["Kevin M. Lahey" <[e-mail]>]

In message <Roam.SIMC.2.0.6.902971866.24899.kcpoon@jurassic>Kacheong.Poon@Eng.S
un.COM writes
>> Most BSD implementation (at least Free/NetBSD) don't have the problem, 
>> because they simply send an ACK every second packet, no matter how big
>> the packets were (this means BSD will send more acks than linux in some
>> situations)
>
>I just got the latest FreeBSD source.  The following part of code has not
>changed.
>[...]
>Can you point me to where FreeBSD forces acking every second segment,
>regardless of the size?

NetBSD does it.  It seemed like a reasonable compromise to me[*];
it avoids assuming that the MSS is the same in both directions,
and it avoids adding extra state and complexity to the system.
In the case of tiny X events or something similar, this results
in extra ACKs.  Are these going to be a significant problem,
perhaps with a grossly assymetric link?

Are there any clocking or other advantages to having frequent ACKs?

Thanks,

Kevin
[e-mail]
-----
[*]  I didn't put it in, so I can't take credit for it, but I certainly
thought it is a Good Thing[tm]...