On 30/12/2016 05:27, John Levine wrote:
... However, my impression is that we
are seeing increasing ISP concentration (except, maybe, close to
the edges of the network, where it makes little difference) and
less of that traditional type of multihoming.
There's tons of multihoming. Every medium sized or larger business
wants multiple upstreams for reliability. They typically get a chunk
of PA IPv4 addresses from each upstream.
This is a big reason why providers don't implement BCP38. A customer
has one block of addresses from provider A and another from provider
B. In general each provider only knows about its own address block,
but the traffic comes from both blocks, and the customers get rather
annoyed if a provider doesn't accept their traffic. ("If you don't
want our $20K/month, we're sure we can find someone else who does.")
Trying to keep track of what customer has what block of someone else's
address space is hopeless, so they just turn off the filters for the
multihomed customers.
This is of course a place where v6 wins, since the customer can
get their own block of PI space,
True, for large sites, but this solution doesn't scale to tens of millions
of customers. For that market, running with several PA prefixes is the answer
(not a problem, although we're still tuning, e.g. RFC 8028).
but then there's all those other
v6 deployment problems.
I'm not saying that those problems are FUD, but they are surmountable. They're
just different from all those v4 deployment problems that people solved some
years ago.
As has been said often enough, users need an incentive to invest in solving
those problems. It will come. Much more slowly than we expected in 1994, but
it will come. I see no reason to panic.
Brian