Re: [lwip-users] performance

[Sam Jansen]

K.J. Mansley wrote:
> But that updates the frequency with which it is checked, not the value
> of the timeout itself?  i.e. It will be more accurately 500ms, but still
> 500ms.

I'm not quite sure what you are getting at. Are you saying the minimum 
bound for an RTO is 500ms or the calculation of the RTO itself will 
result in 500ms?

RFC 2988 actually says,

    Whenever RTO is computed, if it is less than 1 second then the
    RTO SHOULD be rounded up to 1 second.

This is only 'SHOULD', not 'MUST' and in practice this doesn't happen as 
far as I can see: I have a trace here that I measured on our text 
network where Linux 2.4.26 retransmits after around 250ms (this was on a 
fairly low latency link).

As far as the computation goes, there is nothing specifying anything to 
do with 500ms. I can't remember the formula off the top of my head, but 
I imagine lwip implements what is in RFC2988. It's fairly short and 
sweet, see:

http://www.zvon.org/tmRFC/RFC2988/Output/chapter2.html

--
Sam Jansen                                           address@hidden
Wand Network Research Group             http://www.wand.net.nz/~stj2