RE: WG ACTION: 2 weeks to discuss [LL62] Do not space probes randomly

I agree with Christian; randomizing the successive probes may be
superstition ("voodoo" seems a little harsh), but it also is common practice
and I don't think it hurts to specify it here.

BTW, are we assuming the link layer takes care of collisions or should there
be some explicit provision for retransmitting if one of the probes suffers a
collision?

- Ralph

At 04:45 PM 5/4/2004 -0700, Christian Huitema wrote:
> > [LL62]
> >
> > Description of Issue:         Do not space probes randomly
> >
> > ...
> >
> > Why space probes randomly? I have still not seen any credible
> > justification for this, neither credible theoretical analysis, nor
> > experimental data. This text was introduced because of a
>misunderstanding
> > of something Christian Huitema said a year ago, and it's been in the
> > document ever since. An initial random delay is justified. Further
> > randomness is just voodoo.
> >
> > [Erik]
> >
> > This was debated frequently in the WG over the past year and in each
> > case the delay was kept in the document. If you really believe that
>the
> > current text is not reasonable, please contact Christian and see if he
> > agrees with you.  As I recall, the WG agreed with Christian, who
> > supported this text and we ended with rough consensus, with you in a
> > dissenting position. See LL31.
>
>The really important part is the initial randomization. It is then
>important that each successive trial be space by a sufficient delay. But
>it is also important to randomize the second delay to avoid "persistent
>bad luck", i.e. the situation where N hosts pick the same initial delay
>and then all repeat all of their NUM_PROBE packets at the same time.
>Since you need to have a random number generation ready in any case the
>implementation overhead is negligible, so why not do the right thing?
>
>-- Christian Huitema