Re: [RRG] Consensus check: renumbering - all networks need not be treated the same

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

Hi Tony,

Thanks for clarifying the assumptions on which the current poll
question is based:

> |Still, what kind of site does the question refer to?
> |
> |  A site with a single PA IP address such as my DSL service?
> |
> |  A large PA site - hundreds, thousands of addresses perhaps?
> |
> |  A small PI site - whatever that means?
> |
> |  A corporation or university PI site?
> 
> E) All of the above.

You make two statements:

> Any reasonable routing architecture has to deal meaningfully with all of
> these types of sites.  

I agree with this and the first part of your second statement:

> Differentiating between site types is somewhat difficult, as
> inevitably there is some benefit one way or another

but not this:

> and it becomes politically intractable to maintain effective
> separation.


Your current poll question seems to be based on the assumption that
it would be impossible (costly to the point of exceeding all
possible benefits) to distinguish between different kinds of
existing networks when considering the changes each network would
need to make in order to adopt the scalable routing solution.

Perhaps you could put that up as a question for consensus.  I
disagree with this position.

If people agree with this position, then they would also believe
that we can't make different renumbering requirements for any
networks.  For instance, the scalable routing scheme could not
require me to renumber my one IP address DSL-connected home network
whilst not requiring the university a few km away to renumber its
network.

Ideally, no-one would have to renumber.  However, this is a desire,
not a need.  For instance, I don't mind renumbering once - and I
think many smaller networks would be happy to renumber once in order
to get the lasting benefits of the new kind of address space.

I think that if we restrict our scalable routing scheme to meet this
desire (of no renumbering for any networks, or for not making any
distinction between types and sizes of network regarding the
sacrifices each would need to make) as if it were an absolute need,
there will be unacceptable costs, restrictions etc. which would make
the resulting scheme impractical or at least much less efficient and
attractive.


I wrote about this earlier:

  Renumbering once might be OK when converting to Scalable PI (SPI
  space)   http://psg.com/lists/rrg/2008/msg02355.html

I think the scheme should expect anyone with PA space to renumber
once.  I can't imagine any practical method of managing little
segments of space in the middle of existing provider networks via
the new scalable routing scheme in a manner which scales, leads to
optimal path lengths etc.

I think that networks with PI space should be able to take one of
two courses:

1 - Keep their PI space and administer it as part of the new
    addressing scheme - turn it into what I call Scalable PI
    (SPI) space.

2 - Relinquish their current PI space and adopt SPI space
    administered by someone else.

I figure the larger PI networks would have the resources and the
motivation (including not having to renumber) to happily choose
option 1 and that many smaller PI networks would be happy to take
option 2.

  - 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