Date: Wed, 22 Aug 2001 22:08:10 -0700 (PDT) From: singh <singh@pdx.edu> To: Dave Zarzycki <zarzycki@FreeBSD.ORG> Cc: Harkirat Singh <singh@pdx.edu>, Alfred Perlstein <bright@mu.org>, freebsd-net@FreeBSD.ORG Subject: Re: RFC: SACK/FACK patch port to Current Message-ID: <998543289.3b848fba02294@webmail.pdx.edu> In-Reply-To: <Pine.LNX.4.33.0108221958230.19146-100000@bonk.apple.com> References: <Pine.LNX.4.33.0108221958230.19146-100000@bonk.apple.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello All!
I would like to answer your queries as follows:
SACK Model
----------
I looked at PSC web site for SACK/FACK patch for FreeBSD, the latest patch I
could found was for Release2.2.5 (by Luigi Rizzo). After that I surfed web for
any latest information for SACK patch. I even wrote a mail to Sally Floyd,
Luigi and few others about any information on SACK developement in FreeBSD, I
got a reply from Kenjiro Cho (who has worked on ALTQ, ECN, RED etc.) that in
future FreeBSD SACK/FACK would be based on the work (in BSDI) by Hari
Balakrishnan, Tom Henderson and Venkat Padmanabhan, as part of the Daedalus
research group at the University of California, Berkeley
(http://daedalus.cs.berkeley.edu). Please also look at
http://www.aciri.org/floyd/newreno-questions.html.
Then I looked at OpenBSD2.9 and found that it closely follows BSDI and
OpenBSD2.9's SACK and FACK developement is in line with the work done by
Berkeley guys.
I even observed that FreeBSD4.3 adopts to NewReno algorith which is a
suggestion in RFC-2582 (which talks about NewReno, SACK and FACK), for clients
who can not have SACK/FACK, new reno will alleviate the problem of duplicate
acks in Fast Recovery stage and partial ack is a better solution as comapre to
reno algorithm.
In view of above facts I developed SACK/FACK patch in line with OpenBSD2.9 and
BSDI.
Enabling SACK/FACK
-----------------
I used a sysctl to enable and disable SACK, I defined two options "SACK" and
FACK" and added them in kernel option files. I thought that various combination
of SACK and FACK would be as SACK alone, SACK & FACK. FACK alone will not be
used as RFC says that's why I did not add sysctl for it.
Why I post it
-------------
My main aim was to help community as I learned a lot from FreeBSD Newsgroup and
wanted to help people out there so that someone in need of SACK/FACK should not
suffer as I did.
I hope you will find my reply helpfull.
Regards,
Harkirat Singh
Quoting Dave Zarzycki <zarzycki@FreeBSD.ORG>:
> On Wed, 22 Aug 2001, Alfred Perlstein wrote:
>
> > * Dave Zarzycki <zarzycki@freebsd.org> [010822 19:23] wrote:
> > > Attached and tested.
> > >
> > > I'd like to merge this in unless anybody objects.
> >
> > I don't object, I'm just wondering if you know could discuss
> > the implementation here versus the other ones avaiable at:
> >
> > http://www.psc.edu/networking/all_sack.html
> >
> > About the strength/weaknesses of those methods versus this one.
> >
> > The one developed at PSC looks like the most advanced one.
> >
> > Is this work derived from any of those?
>
> I simply ported Harkirat Singh earlier patch from this week.
>
> Harkirat, can you please comment on the above questions?
>
> davez
>
>
> Dave Zarzycki
> Darwin & Mac OS X
> Apple Computer, Inc.
>
>
>
>
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-net" in the body of the message
>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?998543289.3b848fba02294>