ietf
[Top] [All Lists]

Re: multihoming, was IPv10

2016-12-29 21:49:20


--On Thursday, December 29, 2016 16:27 +0000 John Levine
<johnl(_at_)taugh(_dot_)com> 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.

John,

See Brian's comments and some others for responses to what I
think was intended to be your main point.  However, I picked my
words carefully.  I said and meant "traditional type of
multihoming" and, more important in the previous sentence, "many
hosts and sites that are multihomed in the traditional sense of
advertising one set of endpoint addresses to the network and
letting the routing system sort things out".   That is, very
specifically, one address per host, advertised to multiple ISPs/
networks/ paths and not "a chunk of PA IPv4 addresses from each
upstream".  

You are talking about a different --architecturally different--
approach, one that either requires endpoint hosts to manage
different addresses and the routing options associated with each
or that requires NAT arrangements that map one-many or many-many
rather than many-one or one-one.   

I'm not going to argue that one approach is better or worse than
the other.  In particular, unless one is large enough enterprise
to arrange for PI space and for multiple providers to route it,
the traditional multihoming architecture is effectively dead.
But that is another change in the architecture of the network as
it actually functions that we are, to at least some extent,
pretending is not happening.

best,
    john