[v6ops] Re: Dynamic addresses

[Marco Moock <[e-mail]>]

Am Tue, 13 Aug 2024 12:15:06 +0530
schrieb Daryll Swer <[e-mail]>:

> What do you mean by RA prefix lifetime matching PD lease time = no
> problems?

If implemented correctly, there shouldn't be any problems.

> If ISP-side, PD lease time = 24 hours, and RA prefix lifetime = 24
> hours on CE side, and the genius ISP does what it does best, and
> flips the prefixes within <24 hours of elapsed time for the session,
> either due to scripting on their systems that does PPPoE or DHCP
> disconnect per 6-12 hours (yes, there are many ISPs doing this
> globally), or as we discussed on this thread, power issues (varies
> wildly by environments and economies), the hosts on the LAN side of
> the CE will still have old prefixes with original 24 hour pref.
> lifetime value.

They the ISP is the problem because the prefix doesn't have the
lifetime they specify.

> > A router can also extend that to avoid connection interrupts.  
> 
> What do you mean by this?

According to the DHCPv6 RFC, it is possible to extend the lifetime of
PD.

|Each address or delegated prefix assigned to the client has
|associated preferred and valid lifetimes specified by the server.  To
|request an extension of the lifetimes assigned to an address or
|delegated prefix, the client sends a Renew message to the server.

If the ISP implements that properly, it should work, but as you said,
strange ISPs exist.

> As I originally said, this (dynamic prefixes) also makes it virtually
> impossible for the home-user to even host basic SSH to their devices
> on the LAN or even the router itself.

If they are customers of bad ISPs, this is indeed the case.

> I thought we are trying to restore the end-to-end/peer-to-peer
> principle with IPv6, not further push down NAT66/NPTv6 to combat 24/7
> renumbering of the ISP with ULAs and the like. In IPv4, changing IP
> /32 addresses were masked (literally) with masquerade.

e2e exists again in most cases.

> And surely, I hope, nobody is suggesting the folks who authored
> BCOP-690 didn't do their homework. The BCOP clearly explains WHY
> dynamic prefixes == harmful:
> https://www.ripe.net/publications/docs/ripe-690/#5-2--why-non-persistent-assignments-are-considered-harmful

For that case the DHCPv6 PD server could remember that the old prefix
is still valid. I dunno if such implementations exist, but technically
it would be possible.

> Even if we ignore ALL the technical issues with dynamic prefixes,
> dynamic prefixes only increases costs and complexity for law
> enforcement compliance on the ISP-side anyway. In short,
> persistent/static ia_pd is a win-win for both sides (ISP and CE
> owner).

Of course. But if you provides fixed net per customer, you can't charge
an additional fee for fixed net anymore. This is what many ISPs want to
do and already exist with IPv4 - often reasoned by questionable
arguments.