|
|
| -----Original Message-----
| From: Alexandru Petrescu [mailto:Alexandru.Petrescu@xxxxxxxxxxxx]
| Sent: Monday, January 10, 2005 5:14 PM
| To: Pascal Thubert (pthubert)
| Cc: nemo
| Subject: Re: [nemo] RE: about the Home Network Models WG item
|
| Pascal Thubert (pthubert) wrote:
| > | HA proxying for all addresses in all
| > | MNPs in its BC is something that can not be implemented. Why? At
| > least
| > | because proxy NA is sent periodically, becoming an overload. Am I
| > | missing something?
| >
| > Note that I'm not a promoter of that model from the start :) I agree
| > that the HA can not send the async NA for all the range of suffixes
at
| > Home. It should just respond to NS. We can compare this with Dave
| > Thaler's work on spit networks. Alternatively, you can bar coming
Home
| > in Aggregated. Anyway, the process is nit mandated by NEMO or any
other
| > standard, so it would be a product specific support to get there.
|
| If compared to other works (split networks) then it should be
referenced
| and made work with that model. One thousand more lines make sure it's
| exactly the other work, is it worth the effort?
|
| For hosts to autoconfigure addresses based on each MNP - who would
send
| RAs for each MNP on the home link?
|
| > | What does "HA protects the MNP space from autoconfiguration by
| > | uncontrolled visitors on the Home Link" mean?
| >
| > MR binds MNP. Then, a node on the Home Link autoconfs an address
from
| > MNP. HA should defend the address at DAD time. The trouble with
| > aggregated is that the whole aggregation is supposedly on link. Do
you
| > have a rewording in mind?
|
| I don't have a rewording in mind, sorry. My only wording is that it
| can't scale. You seem to say the same thing, so why re-wording it
when
| we could simply remove it.
Well, remove what? The whole aggregated model? The support of coming
Home in aggregated? I agree I can not specify MNP level proxying and
this is an informational draft anyway. All I can say is that if you wish
to do this (nodes at home connect to nodes on MNPs), than you need your
implementation to do that (HA ND proxies for nodes on MNP). I agree I
cannot say how this can be done.
Personally, I believe it could be done, by passively answering NS. But
I'm not 100% sure there's no caveat in the process :( I favor the
extended home network vs. the aggregated.
Pascal
|
|