Re: A common problem with SLAAC in "renumbering" scenarios

[Jan Zorz - Go6 <[e-mail]>]

On 14/02/2019 12:07, Nick Hilliard wrote:
> JORDI PALET MARTINEZ wrote on 14/02/2019 10:21:
>> Why not? This is already happening in IPv4. I've got from a major
>> operator in Spain (and other operators in other countries do the
>> same), the same IPv4 address for decades! I'm talking about
>> residential services, with both DSL and GPON. Not to mention business
>> services were you actually pay for that static address a monthly
>> fee.
> 
> The ietf isn't going to get operator buy-in for mandating static v6 
> prefix assignments.

This is clear and understandable. Do you think they would buy-in the 
advice, that they need to honor the lifetime of a prefix to same DUID 
and give the same PD until it expires, unless there is an urgent network 
design change (moving customers to different BNG, emergency re-route 
because of outage, etc...) and at the same time make sure that CPE sends 
out RA packets with decrementing lifetime on LAN side, according to 
lifetime of PD?

Because at it stands now (if you are not the one with home-brew CPE 
where you fixed this with persistent storage of PD across reboots) - the 
only option that actually works on the field is to give users persistent 
prefixes and that's a fact that operators at large agreed about in 
RIPE-690 document.

So, at the end of a day, it may be in operators interest that we somehow 
fix the whole spaghetti chain between ISP and customer and give the 
advice how to make all 3 points in the network more robust and better 
cooperating between each other when it comes to IPv6...

Cheers, Jan