Re: About Push Model in draft-ietf-6man-addr-select-considerations-02

[Arifumi Matsumoto <[e-mail]>]

Hi,

thank you for your comments.

2010/7/20 Fortune HUANG <[e-mail]>:
> Hi Tim,
>
> In section 7.3 of draft-ietf-6man-addr-select-considerations-02, the second
> paragraph reads:
> "It may of course be possible to piggy back policy information to a host in
> a Router Advertisement message, though initial consensus seems to be that
> this is a less attractive approach."
>
> I think it would be more convincing if there were some text about the reason
> why the push model is a less attractive approach. This would also give
> people a chance to validate this initial consensue.

I remember one of the reasons was its space limitation.
Due to its nature, the distribution policy can be big, e.g. in such a site that
has been allocated a lot of address blocks.

Another reason was that this kind of information is not so basic that should
be carried in RA. It means, this information is very basic and essential in
some environment, but in another environment it can be additional information
for traffic optimiztion, and so on.

Regarding frequent policy update, RA should have some benefits.
DHCPv6 has reconfigure mechanism, but it is not widely used/implemented yet.

It might be good to have both ways.

> Besides, I would be grateful if you could let me know why the second
> sentence of the paragraph has been removed, because I once thought this
> might be a possible solution. The removed sentence reads:
> "However, we may find that RAs may be a good place to indicate whether a
> default policy is in place or not, to avoid hosts requesting non-existent
> updates via DHCPv6."

On this point, we thought the O-bit in RA should be available for this purpose.

Kindest regards,