[Iccrg] Question on RFC 2988 - TCP Retransmission timer

[Lachlan Andrew]

Greetings Ian,

Since you've had no replies, here's my $0.02 worth.

I haven't seen any specific research on that.

Linux currently uses much higher precision timestamps than the old BSD
implementations to which the RFC refers, so it shouldn't really be an
issue.

The fact that Vegas works in Linux also shows that its timestamping
doesn't have systematic errors that would require a more conservative
timeout.

I agree that 1s is a terribly high minimum.  It looks like time for an
update to RFC 2988 :)

Cheers,
Lachlan

On 29/08/2007, Ian McDonald <ian.mcdonald at jandi.co.nz> wrote:
> After some discussion on the Linux networking list I thought I'd ask
> the question here.
>
> In RFC 2988 Section 2.4 says:
>    (2.4) Whenever RTO is computed, if it is less than 1 second then the
>          RTO SHOULD be rounded up to 1 second.
>
>          Traditionally, TCP implementations use coarse grain clocks to
>          measure the RTT and trigger the RTO, which imposes a large
>          minimum value on the RTO.  Research suggests that a large
>          minimum RTO is needed to keep TCP conservative and avoid
>          spurious retransmissions [AP99].  Therefore, this
>          specification requires a large minimum RTO as a conservative
>          approach, while at the same time acknowledging that at some
>          future point, research may show that a smaller minimum RTO is
>          acceptable or superior.
>
> Given that Linux, BSD etc use 200 milliseconds, not 1 second I am
> wondering whether there has in fact been any research done as
> mentioned in last sentence. It seems a very high timeout especially on
> two locally connected devices.


-- 
Lachlan Andrew  Dept of Computer Science, Caltech
1200 E California Blvd, Mail Code 256-80, Pasadena CA 91125, USA
Phone: +1 (626) 395-8820    Fax: +1 (626) 568-3603