ipv6 and failover

[Bjørn Mork]

sthaug at nethelp.no writes:

>> I was afraid for that answer :-(
>> 
>> Are there in the meantime any work-arounds to have some kind of a redundancy, how are Internet providers solving this problem, or aren$,1ryt they?
>
> There are no solutions today (that I know of) if you need customers to
> *keep the same IPv6 addresses* in a failover scenario.

If you are able to pre-allocate the addresses to each client (i.e. you
know the identity of all clients), then you can share/replicate the same
database to as many independent DHCPv6 servers as you like.

And if you don't know the identity of the clients, then then the next
paragraph will likely apply:

> If you *don't* need the customers to keep the same addresses: Simply
> configure two (or more) DHCPv6 servers with non-overlapping address
> pools. IPv6 addresses are not a scarce resource...

Exactly. You don't need failover unless you want to share a dynamically
managed address pool with fewer addresses than clients.  And why would
you want to do that with IPv6?

Take into account the necessary complexity of a failover solution, and I
think you have the explanation why this isn't implememted.  And IMHO
probably never will be.



Bjørn