ticso at cicely7.cicely.de
Thu Feb 14 13:38:04 CET 2013
On Thu, Feb 14, 2013 at 08:59:37AM +0000, Brian E Carpenter wrote:
> On 13/02/2013 17:14, Bernd Walter wrote:
> > My hope was that with IPv6 the return of static addresses will come back
> > again the way I was used from the early days when I started using the
> > internet.
> Not a hope, the design point was exactly the opposite - that all prefixes
> should aggregate as much as possible, which means prefix allocation
> should be dynamic.
I know, but this won't exclude each other.
Aggregating at BGP layer - of course, absolutely required and possible.
ISPs can get a prefix big enough to live with a single allocation for a
ISPs also get enough IPs to aggregate one route per hosting location,
so their own routing tables can be small.
But a customer dialing with DSL is something else.
He is not physically moving and at the access router he he is connected
to needs a routing entry for his prefix anyway.
Of course it means that the customer isn't always handled by the same
access router, but that's the only point where aggregation won't work
and only because they handle customer in an IPv4 way.
If the customer would always be connected to the same access router
than aggregation would work perfectly with static IPs.
This problem is not unsolveable, by either reconsidering the access router
mechanisms or by additional space for routing tables.
If a customer is physically moving then it is OK to hand him a new
They offer it for commercial customers - at a higher price.
The higher price is basicly Ok because it means a bigger local routing
table, which needs to be paid for, or they have to tag the customer to
a single access router, which then again can have a single aggregated
It is not Ok that this option is only available to business customers.
Most customers, even business, won't need this - hand them out dynamic
prefixes to reduce cost, but offer them a static PA prefix if they have
a use case.
Many even want to have dynamic because they believe in the dynamic
means privacy hoax.
> > With IPv6 there is an implication because it is not only the single
> > public address which change - the whole internal network gets renumered
> > on redial-in.
> Yes, but what's new? When I reboot my domestic ADSL box, it renumbers
> all hosts. They are always within 192.168.1.0/24 as it happens, but as
> far as my laptop is concerned, it gets a new address each time.
When your ADSL box renumbers your local network on reboot, then you
have to live with it or replace it with something better.
It is completely your decision - even if the access router is owned by
your provider there are options for you to handle this e.g. with an
intermediate box, which you might do anyway to have a decend firewall,
or a fully configureable DHCP server.
> As far as I can see, ULAs are an improvement on this.
Yes it is.
My default style is multihomed as it reflects my typical customer, but
I also see NPTv6 as a good option for other use cases.
Both options have drawbacks.
It is also good if you don't want to pay for the static prefix option
and can live with this, or as named to use NPTv6 to simplify provider
change (a managed decision to renumber) instead of using costly PI with
the high impact on global routing tables.
It is not an option for every case.
B.Walter <bernd at bwct.de> http://www.bwct.de
Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.
More information about the ipv6-ops