Re: W/RTT verification, linux tcp buffers behaviour

From: Al Boldi
Date: Fri May 19 2006 - 07:40:22 EST


John Heffner wrote:
> Al Boldi wrote:
> > Which module would you suggest for a non-congested network?
>
> "It depends." There are quite a few papers out there on this. It's
> currently an area of active research.

Which one/s do you use?

> > Or is there a module that allows for a=(n+x)*MSS where x is tunable?
>
> You could write one. I don't know that you would want to...

I could probably modify one. Could you point me to the part that needs
changing?

> >> If you are actually CPU limited, on an SMP system you can also get
> >> better performance with multiple connections, since they can be handled
> >> concurrently by multiple CPU's.
> >
> > It seems that CPU cycles are related to latency, which is related to
> > bandwidth-delay. So a faster CPU should yield higher thruput, which it
> > does, but this would mean, that thruput is dependent on system load as
> > that affects latency.
> >
> > Is there a module that takes this fluctuation into account, to yield a
> > constant flow?
>
> Sorry, not following. CPU cycles only add to latency if you don't have
> any. If you don't have spare CPU cycles, you're not going to be sending
> any faster!

Sure, but faster sending is only one aspect of higher total thruput, the
other aspect being constant/non-fluctuating flow, as TCP thruput is rather
sensitive to fluctuating latencies.

So, is there a module that takes this fluctuation into account, to yield a
constant flow?

Or do I have to write that one too :)

Thanks!

--
Al

-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html