Re: [RRG] IPv4/6/ngng or IPv4-map-encap then IPngng

[Robin Whittle <rw@firstpr.com.au>]

Hi Per,

You wrote:

>> For the RIRs to seriously consider the possibility I am
>> suggesting, they would need to have a good understanding of the
>> possibilities of map-encap schemes.
> 
> From where do you get the idea that these groups are so
> completely disconnected? There's always been disputes between
> ISP-engineers with a preference for practical working solutions
> and idealist scientists. Still, leading engineers in the provider
> industry who are active participants in governance-communities
> are closely following these developments as subscribers on these
> lists. Your assumptions of ignorance is offending a lot of
> people.

I don't mean to offend anyone.

I don't know what the membership of the RRG list is.  Mailing lists
are weird in that there could be hundreds of people reading
everything intently and even discussing it privately amongst
themselves.  If  they never write to the list, then those who do
write might never know they exist.

My impression is that only a subset of active RRG members support
map-encap at all.  Of them few if any have expressed serious support
for Ivip.  I have collaborated with a lurker on a paper about
mobility - yet to be finalised - so I know at least one RRG member
has confidence in my proposal.

Your direct knowledge of who reads this list is presumably better
than my lack of knowledge.  If there are folks keeping up with all
this, and who think IPv4 map-encap is worth considering, then I
suggest that now would be a good time to de-lurk and join the debate.


>> Yet these schemes - with the exception of the LISP prototype
>> code - are vapourware and hardly known outside the RRG.  Even
>> within the RRG not many people have much faith in them, and
>> each of us map-encap developers has less faith in the other 
>> systems than in our own!
> 
> If the RIR communities are so ignorant then why have such
> solutions been discussed both on and off-stage at community
> conferences for years?

While map-encap in principle has been suggested for years, as far as
I know, I was the first to propose a form of map-encap which
supported backwards compatibility for full connectivity (with
portability and multihoming) for packets to and from non-upgraded
networks.  In proposing this, I think I was the first to show that
one could create independent address prefixes for hundreds or
thousands of end-user networks, each portable and multihomable,
within a single shorter prefix which constituted a single BGP
advertised prefix.

That was my way of improving LISP, to create the beginnings of my
Ivip proposal, on 2007-06-15:

  http://www.ietf.org/mail-archive/web/ram/current/msg01518.html

Before that LISP had no way of being introduced with multihoming of
communications from non-upgraded networks.  (Late last year LISP
became able to do this, by adopting Proxy Tunnel Routers which do
the same job as my poorly named "anycast ITRs in the core" - now
"Open ITRs in the DFZ".)

The way I am suggesting IPv4 space be finely sliced up and used as
PI space for large numbers of end-user networks, without worsening
the routing scaling problem, AFAIK has its roots in my proposal.
Maybe the LISP team don't think their system could or should be used
this way, but I think it can and should be.

Maybe such proposals have been floating around for longer than this,
but it was my impression they are about a year old.  Maybe lots of
RIR - ISP people are totally up to speed on all this.  I hope so.


>> I imagine there would be a curved relationship between the
>> price per IP address and the number of IP addresses in the
>> prefix which end-users want.
>> 
>> I think you could find hundreds of thousands of end-users who
>> would pay, say $20 a year per portable, multihomable IP
>> address, for small amounts such as 1, 2, 8, 32 or so.
> 
> A problem with this is that micro-allocations only make up a
> small portion of the annual global consumption. If the big
> consumers are pushed into such prices/address you'll very soon
> see considerable investments going towards alternative solutions.

I agree that if the big ISPs can't get the space they need they will
be highly motivated to trying to figure out how to sell IPv6-only
services to home and SOHO customers, each of whom chew an IPv4
address each at present.

My understanding is that the routing scaling problem is driven
largely by end-user networks needing PI space for multihoming and
TE.  As far as I know, a lot of those end-user networks could get by
happily with less space than they currently get from RIRs, if a
map-encap system was working as I intend.  Then, there are a much
larger number of other end-user networks who currently don't have
the resources to get BGP-managed PI space who could also get these
smaller, cheaper, scalable slices of map-encap managed PI space.

Map-encap provides a drastically better form of management for PI
space than the current BGP system, which for various reasons which
are unlikely to change doesn't slice up space into smaller chunks
than 256 IP addresses.

Map-encap provides much finer control of space, without burdening
the BGP control plane - and so will surely enable a much better
utilization of IPv4 space.

I can't quantify this.

However, given the apparently low actual usage rates of IPv4
addresses, such as less than less than 20% on most advertised prefixes:

http://www.firstpr.com.au/ip/host-density-per-prefix/
http://www.firstpr.com.au/ip/host-density-per-prefix/all-graphs/
http://www.isi.edu/~johnh/PAPERS/Heidemann08a.pdf

even allowing for ping probes seriously under counting actual usage,
and with 3.7B addresses which could be advertised when 1.7B are
currently advertised, I think there is tremendous scope for more
IPv4 addresses actually being used in the next 10 to 15 years.

 - Robin


--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg