[arin-discuss] Anyone play with IPv6 on the RV4000?

[Dean Anderson]

Hi, Ted,

On Thu, 29 May 2008, Ted Mittelstaedt wrote:

> I picked up a little consumer ethernet-to-ethernet router the
> other day for about $120 USD and surprised that is has IPv6
> capability in it.  It's a Linksys model RVS4000.  

Linksys runs mostly embedded linux. Linux has had some kind of IPv6
support for a long time, though IPv6 is usually disabled in the embedded
versions because of space constraints.  Its rather interesting that
linksys assumes there will be IPv6-only clients, since there are no such
beasts at this time.

> Does the existence of this at that price point make the "we gotta
> extend the life of IPv4" argument a moot issue?

The 'life' of IPv4 is not at issue. We are trying to extend the time
before the complete exhaustion of unassigned IPv4 address space.  
Avoiding exhaustion is just good management and responsible policy. But
even in the presence of bad management, IPv4 will continue to 'live' for
quite a long while after there is no more unassigned IP address space.



BTW, someone asked me if I had any ideas on an alternative to IPv6, if
it fails.  The ultra short answer is yes: OSI CLNS is currently
implemented in just about every router vendor which supports
IS-IS---which is every major brand and most minor brands.  All that
remains is an OSI CLNS stack for client platforms, and/or an IPv4-CLNS
proxy. This is actually much easier than implementing IPv6 on those
platforms since there are already longtime stable implementations of the
OSI CLNS stack. The other advantage of OSI is that there is a /variable
length/ addressing scheme so that packets can be made much smaller than
IPv6 packets, and the address space can grow in size without changes to
the software.  Indeed, in retrospect, choosing IPv4 was probably a
mistake because of the fixed address size limitations. But of course, at
the time people probably didn't anticipate that address space exhaustion
would ever be a problem.

And keep in mind also that IS-IS is also superior to OSPF in that every
OSPF router in the network must be upgraded to support a new link type.
By contrast IS-IS will just pass this information around transparently,
so when you add a new link type to the (IS-IS routed) network, only
those routers with the that new linktype need to be upgraded. Very nice.
IS-IS can route anything, including IPv4 and IPv6, and of course OSI.  
Many large sites use IS-IS in place of OSPF to route their IPv4
interior. 

BGP has similar flaws as OSPF. OSI has an exterior protocol called IDRP,
which appears similarly to IS-IS, to be able to route IPv4 as well as
OSI. I anticipate (but have no data to confirm), that IDRP can be setup
similar to a BGP route reflector (ala zebra/quagga) to exchange exterior
OSI and IPv4 routes amoung peers.  Does anyone want to work on that?

Have a good weekend,

		--Dean

-- 
Av8 Internet   Prepared to pay a premium for better service?
www.av8.net         faster, more reliable, better service
617 344 9000