Re: Guidelines for Numbering IPv6 Point-to-Point Links and Easing the Addressing Plans

[JORDI PALET MARTINEZ <jordi.palet@consulintel.es>]

Hi Vincent,

See below, in-line.

Regards,
Jordi




> De: Vincent Jardin <vincent.jardin@6wind.com>
> Organización: http://www.ipv6.6WIND.com/
> Responder a: <owner-v6ops@ops.ietf.org>
> Fecha: Wed, 01 Mar 2006 11:30:38 +0100
> Para: <jordi.palet@consulintel.es>
> CC: "v6ops@ops.ietf.org" <v6ops@ops.ietf.org>
> Asunto: Re: Guidelines for Numbering IPv6 Point-to-Point Links and Easing the
> Addressing Plans
> 
> Hi,
> 
> Please, this draft should metion:
>   - the cases when there are only link-local addresses
> 
>   - because it is mainly focus on ISPs cases, it should describe too how
> it fits with Prefix Delegation: my recommandation would be not to put
> any addresses ont the point to point interface (there is already
> link-local addresses which are enough for routing protocols).

I disagree regarding using link-local addresses, is very confusing when you
do a traceroute and you can even see the same addresses in different
customers in different links (if de ISP decides to use a fix IID for each
side).

This has been also the feedback that we got from different ISPs.

> 
>   - there are some drawbacks of putting an IPv6 /64 on the provider side,
>        + because you will create loops if redirection is not properly
> implemented on point to point interfaces (I mean a good implementation
> and configuration should never send redirect packets)
>        + because you will loose the support of the anycast address
> sub-network. It means that the LNS or BAS will answer instead of the CPE.
> 
>   - what's about the support of the anycast address sub-network ?

The /64 is in the customer side, not the provider one, which uses /48.

> 
>   - what's about multi-links to a same site ? For instance, we know some
> cases where you have 2 point to point links with ECMPv6 for load
> balancing. Section 4 cannot be applied in that case.

Not sure what do you mean with ECMPv6. I will check it out tomorrow, I'm off
line now.

> 
> FYI and from my experience, for ISPs, it is useless to number the IPv6
> links. Only link local is enough.  From an implementation point of view,
> you just need to add a reject (or blackhole) /48 route for each users
> through each interface. Because you have a same route through many
> interfaces, you will have the benefit of ECMPv6 support.

I think this is what actually I'm proposing then ;-)

> 
> Maybe, it could be useful (and I do doubt), to number only the IPv6
> point to point links that are more used for infrasturcture (GRE, 6in6,
> L2TP/VPN, etc.).
> 
> Regards,
>   Vincent
> 
>>> Hi all,
>>> 
>>> We have submitted before yesterday a new draft, it seems is
>>> still not available at the IETF repository, but meanwhile,
>>> the document can be reached
>>> at:
>>> http://www.consulintel.euro6ix.org/ietf/draft-palet-v6ops-poin
>>>    
>>> 
>> t2point-00.txt
>>  
>> 
>>> Comments welcome !
>>> 
>>>    
>>> 
> 
> 




**********************************************
The IPv6 Portal: http://www.ipv6tf.org

Barcelona 2005 Global IPv6 Summit
Slides available at:
http://www.ipv6-es.com

This electronic message contains information which may be privileged or confidential. The information is intended to be for the use of the individual(s) named above. If you are not the intended recipient be aware that any disclosure, copying, distribution or use of the contents of this information, including attached files, is prohibited.