[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[altq 1225] Re: Random number generator
On Mon, 3 Dec 2001, Luigi Rizzo wrote:
> On Tue, Dec 04, 2001 at 01:13:09AM +0900, Kenjiro Cho wrote:
> > I also did a quick test. The result is good when I use 1,000,000
> > samples (since 1,000 is too small to test stochastic properties).
> not to mention that it is RED we are talking about... drop rates
> need not to be precise, you just need to drop "some" packets and
> pray you have chosen the right set of values (assuming there is
> one) for your traffic mix.
I didn't say RED but RED-like active queue management (its real
name is PI - see On Designing Improved Controllers for AQM Routers
Supporting TCP Flows. Infocom 01.). The claim is that PI's
control-feedback scheme is somewhat sensitive to numerical error.
Furthermore, its update interval is fairly small (a few hundred
milliseconds or less).
On a 10-Mbps link, 1000000 samples would correspond to 1000 seconds.
Maybe, this is an inherent problem for PI because it requires both
fast and accurate random number generator.
In any event, this was why I posted my original message asking for an
altenative random number generator.