(6bone) Ingress filtering (was: asymmetric routing)

[Joel Baker]

On Fri, Feb 01, 2002 at 07:53:01PM +0200, Pekka Savola wrote:
> On Fri, 1 Feb 2002, Matteo Tescione wrote:
> > > Problem with IPv6 multihoming is that IPv4 multihoming is so "easy" and
> > > works quite well.  It may be we can't design a protocol or equivalent that
> > > will handle the scenarios responsibly and as well.
> > Easy???
> 
> Just get prefix A.B.C.0/24 and two ISP's, pay them, configure BGP with
> them and advertise as appropriate.

Indeed. As I've said in other fora: the going rate for aquiring a swamp
/24 on the gray market, the last time I checked, was about $30,000.

> > According to you if ipv4 multihoming is easy, ipv6 multihoming will be
> > impossible...
> > And if ipv6 multihoming will be impossible I suggest to stop experimenting
> > ipv6.
> 
> Who said the (only) goal of IPv6 was multihoming?
> 
> This can be seen as a good thing: it forces everyone to stop the
> irresponsible practise of cluttering global routing table (among others).  
> A drawback is that we don't have anything really concrete to offer for
> site multihoming problem in the place of old practises; there are a few
> proposals though.

You say "cluttering global routing table", I say "protecting my business
investment from being vulnerable to my upstream provider's vagaries". The
Internet is no longer comprised of a bunch of folks who are mostly using
it as a giant experiment and are losing no money for changing over (such
as was the case when IPv4 was adopted, for the most part).

For the record... I strongly favor the adoption of protocols such as SCTP
which can handle the fundamental problem of one machine having multiple
addresses on different networks (which was not actually part of the early
design requirements, except for routing gateways), which work under both
IPv4 and IPv6, and remove something like 90-95% of the use for BGP for any
form of end-customer (IE, non-ISP folks). As has been pointed out to me
privately, there are already folks who clearly are *not* ISPs who have
found it worthwhile to apply for pTLA status, simply to ensure that they
have a globally routable block of space.
-- 
***************************************************************************
Joel Baker                           System Administrator - lightbearer.com
lucifer@lightbearer.com              http://users.lightbearer.com/lucifer/