Re: [v6ops] Fwd: New Version Notification for draft-hilliard-v6ops-host-addr-update-00.txt

[joel jaeggli <[e-mail]>]

On 7/17/17 22:58, Alexandre Petrescu wrote:
> 
> 
> Le 17/07/2017 à 22:33, Ross Chandler a écrit :
>>
>>> On 17 Jul 2017, at 20:30, Ted Lemon <[e-mail]
>>> <mailto:[e-mail]>> wrote:
>>>
>>> What the document actually implies is that individual address
>>> assignment using DHCPv6 is not recommended; instead it is
>>> recommended that each host get a /64.  The only way to do that
>>> right now is with DHCPv6 PD.   So the document is explicitly
>>> recommending the use of DHCPv6.
>>
>> https://tools.ietf.org/html/draft-ietf-v6ops-unique-ipv6-prefix-per-host-06
>>
>>
>>  Section 4 describes a way of assigning a /64 per host using RAs that
>> is already deployed in real networks. No DHCPv6 involved.
> 
> There is something wrong with that draft: it only allows for 64.  Why?
> DHCPv6-PD can delegate other plengths, like 65.

"a Unique IPv6 prefix (currently a /64 prefix)"

seems like a coordination problem. where it would be best not to
constrain the delegated prefix if it assumes that 64bit iids will be
assigned via slaac then /64 is the lower bound, however it could be
shorter e.g. 60 56 48 etc. so it would seem to be possible to delegate
the appropiate sized prefix for that application.

>> https://tools.ietf.org/id/draft-pioxfolks-6man-pio-exclusive-bit-02.txt
>>
>>  Introduces an eXclusive flag to optimize RA for nodes that are
>> exclusive receivers of all traffic to the prefix.
> 
> This is another way of calling RA "Prefix Delegation".
> 
> There are other drafts doing Prefix Delegation with RA, have you
> considered them?
> 
>>> However, if the only DHCP service available is individual address
>>> allocation, then indeed that is not recommended, because it has
>>> serious privacy implications.   And it is not _generally_
>>> recommended that people operate networks that require DHCPv6 static
>>> individual address allocation (IA_NA) for this same reason.   Using
>>> the DHCPv6 privacy profile does mitigate this concern, but still
>>> the best thing to do is just enable SLAAC.
>>>
>>> I don't think these views are particularly controversial in the
>>> IETF. I'm one of the authors of RFC3315, and I agree with this
>>> view.
>>
>> Given the above two drafts DHCPv6-PD to hosts has an uphill struggle
>> to ever achieve widespread deployment.
> 
> I dont think so.
> 
> I think DHCPv6-PD should be trialled at IETF WiFi and then we'll see.
> 
> Alex
> 
> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
>