Re: [v6ops] Fwd: New Version Notification for draft-linkova-v6ops-ipmaclimi-00.txt

[Mark Smith <[e-mail]>]

On Tue, 8 Nov 2022, 23:39 Nick Hilliard, <[e-mail]> wrote:

> Mark Smith wrote on 08/11/2022 12:19:
> > So just using DHCPv6 doesn't inherently ensure this problem won't occur.
>
> In an SLAAC environment, the operator becomes hostage to the policies of
> network clients.  At least in DHCPv6, there is a mechanism for the
> operator to manage this.
>

True.

I'm just making the point that it can't be said that using DHCPv6 rather
than SLAAC naturally prevents this problem, which seemed to be your
assertion.

These problems fundamentally exist because hosts are in a position to
create state in network elements without getting authorisation and
therefore exhaust that state capacity

This is not a new problem, IPv4 ARP caches have always been vulnerable to
state capacity exhaustion.

There should be upper limits on how many resources an individual host can
consume, however those limits need to reasonably allow hosts to do what
they want to do, since the network exists to serve hosts.

I don't have much exposure to hosts that are using many IPv6 addresses at
once, so don't have any rough idea of what is typical to then recommend as
a default minimum where there could be a resource default limit such as
these wifi ND proxies.

However, as RFC1681 seems to be the origin of IPv6 multi-addressing, and it
suggested hosts should be allowed to have somewhere between 64 and 256
concurrent IPv6 addresses.

Regards,
Mark.



> Nick
>