BCP for multisite multihoming

[Per Heldal]

On Mon, 2007-05-21 at 14:43 -0700, David Conrad wrote:
> The IETF is mostly irrelevant in this context.  The folks who have to  
> get over their fear (irrational or not) is the Internet service  
> provider community.


Provider-fear may be part of the problem, but it's not fair to only
point to routing for solutions. At this stage we're really just looking
at band-aids to compensate for the lack of scalable
inter-domain-routing. 

Looking beyond interpretations by network-people I have never found PI
mentioned among the business- or functional-requirements for networking
in end-sites. PI is a possible solution, not a requirement. The vast
majority of businesses have no desire to deal with the complexity of
using PI for portability, redundancy, TE etc. Flexibility similar to
what number-portability for telephones provide in regulated markets is
adequate for most end-sites.

Maybe a part of the solution could be to establish viable alternatives
to PI. There are imho several issues which may have greater impact on
routing-table-growth than RIR- and routing-policies wrt PI. A couple
examples:

1. Ever since the introduction of CIDR the configuration of IP devices
(firewall-rules etc. included) should have separated site-prefix from
the local address, with mechanisms to propagate site-prefixes across the
local network. With IPv6 this becomes even more important. While
v6-advocates preach the elimination of NAT, there is no viable
alternative (except PI) to avoid renumbering-pains. 


2. DNS, a key to portability (similar to number/line-abstraction in
phone networks), must be subject to strict regulation to prevent
DNS-operators from keeping their customers "hostage". One may even have
to consider a fresh start if the current DNS-tree and the associated
domain-industry is corrupted (contractually, financially and morally)
beyond repair.




//per