> -----Mensaje original-----
> De: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]
> Enviado el: viernes, 08 de febrero de 2002 18:29
> Para: marcelo bagnulo
> CC: Alexandru Petrescu; [EMAIL PROTECTED]
> Asunto: Re: Randomness and uniqueness
>
>
> In your previous mail you wrote:
>
> if you do agree with the probablity argument
> but you do not like to implement it, because you do not like it,
> what are the reasons for this dislike?
>
> => the probability argument gives no guarantee.
Nothing can give you full guarantees There is no option that always work,
all the options work fine with a certain probability, The real question is
if the probability that it fails is low enough. And what does low enough
means basically depends on the probablility of failure of the others
elements related. So the question to be answered is if the probability of
colision with random generated address identiers is low enough compared with
the probability of failure of the other devices and protocols in the net.
This is what we are trying to quantify.
It is also important to consider the cost, i.e. most of the times you can
improve a solution (meaning that you can get a lower probability of failure)
but this usually implies aditional costs, the question is if this is worthy.
I mean whether you like it or not you are playing russian roulette :-), DAD
can fail as well as all the other elements in your network
regards, m
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
