IPv6 Pain Experiment

Doug Barton dougb at dougbarton.us
Fri Oct 4 03:14:50 UTC 2019


On 10/3/19 5:35 PM, Masataka Ohta wrote:
> Doug Barton wrote:
> 
>> Not if you configure your services (like DNS) with static 
>> addresses,which as we've already discussed is not only possible, but 
>> easy.
> 
> That's your opinion. But, as Mark Andrews said:
> 
>  > Actually you can do exactly the same thing for glue.

If Mark wants to do that on his network, he can. That doesn't mean that 
you have to.

> I show it not so easy.

No, you've shown that there are ways to do things differently than using 
static addresses for your services.

>  > Please stop spreading FUD regarding this topic.
> 
> Automatic renumbering involving DNS was important design goal
> of IPv6 with reasons.
> 
> Lack of it is still a problem.

If you're talking clients (system using only outbound connections) then 
they will renumber with SLAAC+DHCPv6 the same way that IPv4 clients 
renumber with DHCP now.

If you're talking about services (with inbound connections) then yes, IF 
you ever have to renumber, AND you use static addresses instead of 
SLAAC, you'll have to do them by hand.

But weren't you just arguing that dynamic addresses for services is a 
bad thing? Which way do you want it? Because you can have it either way. 
In fact, you can have it BOTH ways if you want it, depending on the 
service. I find static addresses for services easier myself, but Mark is 
a lot smarter than I am, so I'm perfectly ready to be proven wrong.

Meanwhile, the thing that most people miss about IPv6 is that except in 
edge cases, you never have to renumber. You get a massive address block 
that you can use as long as you pay your bill. The chance that you'll 
have to renumber, ever, is incredibly small.

So, again, stop spreading FUD.

Doug




More information about the NANOG mailing list