Le 25/10/2012 20:25, Brian E Carpenter a écrit :
[...]
I checked the document and it misses, I think, something important.
The first and most naïve request of a prefix by a Requesting Router
should have all bits zero and maybe the prefix length 0, or around
64.
Would this be ok?
Wouldn't the request typically be for a /48 or a /56? If I'm
renumbering an existing network, I will want a prefix that is no
longer than what I already used.
The 'end' router in a Leaf Vehicle would probably need 1 or 2 /64 for
its own in-vehicle links. Or maybe a single /63 to make these 2 /64s
our of it. Or similar. This MR-LV would not sub-delegate to other
vehicles.
On another hand, the router in an Internet Vehicle IV would use
DHCPv6-Prefix-Delegation to obtain an as short prefix as possible (/48,
/56) because it knows it would 'sub-delegate' to other LVs - it has
dedicated interfaces for that.
I wonder whether an expensive BMW will always request a shorter
prefix than a Nissan Versa.
That is an excellent comment.
The IV would have a SIM card with precise billing scheme; existing more
expensive cars have higher subscription plans. One would expect a large
BMW connected to e.g. T-Mobile to be more likely to obtain a shorter
prefix (or several longer prefixes) than a cheaper car.
On another hand, the LV without a SIM would be subject to more flexible
billing plans, like with free-access WiFi but advertisement, and where
the notion of longer or shorter prefix would less be related to the cost
of the vehicle.
I think the ND-PD happening between IV and LV would not put restrictions
on the length of prefix requested or allocated.
Let me add the the size and cost of vehicle is not much related to how
much electronics and computers it has inside. Smaller vehicles are more
and more automated and full of devices. Also, it may be that new small
vehicles coming out these days have much more computers inside than an
older but large and comfortable sedan.
Alex
It also says:
PC_ID: An unique identifier of the Prefix Collection.
The PC_ID MUST be unique among all PC_ID known by the
requesting router.
How can the requesting router provide this in its REQ message?
By guessing?
I think the PC_ID is generated by the requesting router (not by
the delegating router), just as with IA_ID, no?
Ah yes, I simply misread the text, sorry!
Brian
11. Security Considerations
TBD
OK, but since a vehicular network is open to any one of millions
of unmanaged devices, this will need to be *very* convincing,
especially in preventing DOS.
I agree. We currently try to understand these threats, and
especially how they are specific to ND, to prefix delegation and to
new flags we add.
We may add more threat descritpion in the next versions of the
draft.
Alex
Regards Brian Carpenter
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
