Hi Geoff,

Geoff Huston wrote:

I see that the local use address draft has been revised and published as a WG
document.


In section 3.2.2 of the draft, it notes that, in reference to Locally
Assigned  Global IDs that  "the likelihood of conflict is small. "

I had noted in draft-huston-ipv6-local-use-comments-00.txt that  the
likelihood of conflict is given by  the formula:

P = 1 - ((n!) / ((n**d)(n-d)!))

where n = 2**40 and d is the number of random 'draws' from the pool.

The likelihood of conflict exceeds 0.5 after only 1.24 million draws. I'd
contend that this is definitely not "small" as described in the draft.


My disagreement with this calculation is two-fold.

The first issue I have is that these addresses are not meant to be routed
globally, so there is an additional variable that needs to be considered
when calculating P for collisions.  That is, the P above needs to take into
account the probability that two networks trying to use the same prefix are
connected.

My second issue is with the formula itself.  RFC 1889 uses an MD5 hash for
calculating its SSRC Identifier (a 32-bit number).  That document states:

         the probability that two sources independently pick the same value
         can be approximated for large N [20] as 1 - exp(-N**2 / 2**(L+1)).
         For N=1000, the probability is roughly 10**-4.

N represents the number sources (number of entities making a random selection)
and L is the length of the identifier. I see this formula being more representative
of the probability of collision.


Regards,
Brian H.

--------------------------------------------------------------------
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]
--------------------------------------------------------------------

Reply via email to