Re: [tcpm] RFC 1323: Timestamps option

Hi,

On Fri, 26 Jan 2007, David Borman wrote:

> But I totally agree, just feeding in the more frequent samples from                                                           
> the Timestamps into the RTO estimator is not a good thing.                                                                    

Indeed. Naturally, it is possible to adjust the RTO estimation to be able
to deal with receiving estimates more often, or even irregularly. Quite a
few OSes (eg Darwin, I think?) use a RTO timer (and time stamps) that ticks
every half a second, and so effectively rounds up to the nearest 500ms.  I
guess this shows that it doesn't matter too much what you do, as long as
you give a reasonable overestimate.

It is conceivable that people might want to use timestamps for things other
than RTO estimation. Some of the high-speed TCP variants want more RTT
samples (we're playing with one that expects an RTT sample per ACK at the
moment).  I don't think any of the suggested new behaviours would cause
problems with these uses, but it may be worth keeping in mind while
updating RFC 1323.

As an example of a more exotic use of the time stamp option, the most
recent version of FreeBSD's syncookies stashes information in the
timestamp:

        http://lists.freebsd.org/pipermail/freebsd-net/2006-September/011651.html

Gavin

_______________________________________________
tcpm mailing list
tcpm@ietf.org
https://www1.ietf.org/mailman/listinfo/tcpm