Re: draft-ietf-v6ops-addcon-03.txt ... ULAs of shorter-than-/48 and ULA multicast scope matching ...

[Brian E Carpenter <brc@zurich.ibm.com>]

On 2007-03-26 14:35, Tim Chown wrote:
...
> The question here is what text should we put in the addon draft?  We
> would like to have some text for three cases:
>
> a) Site has multiple /48 prefixes (e.g. a /44) and wishes to use ULAs
> b) Site has one /48 and wishes to use ULAs 
> c) Site has a less-than-/48 prefix (e.g. a /56 or /64) and wishes to use ULAs
>
> For a) the discussion was whether to pick 16 'random' /48 ULA prefixes or
> to pick one ULA prefix using 36 random bits rather than 40.

I would recommend the first as being strictly conformant
to the spec, but observe that the second does not create
architectural issues.

> For b) I think we have no issue :)
>
> For c) does the site use a random 48-bit ULA prefix with 0's to the actual
> network boundary, or does it minimise collision likelihood (in particular for
> cases of network merger or overlap) by randomising the bits to the network 
> boundary?

Architecturally, it doesn't matter. The collision issue seems
negligible to me. So I think it *really* doesn't matter. I think
I would just use zeros.

> The simplest text is to use fixed /48 ULAs in all cases.   The text could
> describe the above three cases and state that conclusion and note briefly
> the tradeoffs.

+1

    Brian