Folks,
> James Kempf wrote:
> The DNS is a node identifier, but it currently can't be used by
> the transport layer as a connection endpoint identifier.
Nor by the network layer.
> Brian E Carpenter wrote:
> The reasons why FQDNs are imperfect EIDs have been listed quite
> recently (on one of the IPV6 lists I think).
Indeed. Besides, what we are talking about is scoping, and one of the
things that is totally in scope is to discuss at which layer(s) scoping
shall occur.
IMHO, scoping shall occur at the IP layer. The fact of the matter is (as
this discussion as shown) it seems that scoping is tied to the
identifier. I think that one of the sound discussions we should have is
to identify if indeed this is true. My vote is yes.
Now, _if_ the identifier is moved above the network layer, and if scope
is tied to the identifier as we might or might not decide it is, we
suddenly have a layer violation, because scope is going to be used at
the routing layer (the IP layer) and for the same reason routing must
not use DNS (because DNS needs routing) we will likely bump into the
issue that scope could not use {whatever the identifier in an upper
layer is} because of the same circular reference issue.
So, IMHO the beginning of the decision tree is:
|
v
|
/\
/ \
/ Is \ +----------------------------+
/scope \ | Since we don't have |
/ tied \ | identifiers at another |
/ to the \ yes | layer than IP now, |
\identifier/--->---+ scoping is to be done |
\or not? / | at the network layer. |
\ / | if we ever get identifiers |
\ / | in other layers, situation |
\ / | will be revisited then. |
\/ +----------------------------+
|
v no
|
+----+----------+
| what are the |
| reasons to |
| move scope to |
| another layer |
| than IP? |
+----+----------+
|
| ???
|
Comments?
Michel.
_______________________________________________
Saad mailing list
Saad@ietf.org
https://www1.ietf.org/mailman/listinfo/saad
