RE: [RRG] Consensus? End-user networks need their own portable address space

[Dan Jen <jenster@cs.ucla.edu>]

On Fri, 2008-06-20 at 08:39 -0700, Tony Li wrote:
> Yes, the transition to this is not smooth, but unless we create a new
> namespace, we are effectively endorsing the semantic overload that we
> have
> today and will have to live with it in perpetuity. 

I'm not sure about everybody else, but lately I have had a lot of
trouble determining the constraints of our proposal.  I'll write what I
think is going on, and let me know if I made any mistakes.

We need an architecture that solves the scalability problem in routers,
while simultaneously providing easy multihoming and easy connectivity
change for customer networks.  The architecture must be made compatible
with v6.

For any proposal that requires any change(basically any proposal), there
is an apparent constraint that some driver must be present to ensure
that these changes occur.  I'll call it the 'driver constraint'.  Since
Tony wants a long-term solution, we can wait 30(more?) years for this
driver to appear or grow naturally.  We can even do things in that time
to encourage the growth of drivers.  But it still must be present, so
economic issues ARE relevant to us, correct?  Technical issues are also
relevant, of course.  But at some point, some network operator needs to
be motivated to make the changes necessary for the proposal.

The RRG has not yet been able to agree on what motivations are possible
in the long term (have we even decided on the duration of this 'long
term'?  1000 years?  50 years? 10 years?).  An example of this lack of
agreement is that Bill Herrin just made arguments for the need for PI
space due to some current operational practices.  Tony Li rebutted that
these practices can change over time.  Is it safe to assume that any
proposal requiring these changes must come with some reasoning that
these changes will indeed occur over time?  

Robin suggests that instead of trying to develop a general driver
constraint, we should recommend specific proposals and discuss them.
For any particular proposal, we know the changes proposed.  We can argue
about the feasibility of the changes and evaluate the proposal based on
this as well as the proposal's effectiveness, security, etc.  However,
Tony doesn't think this is a good course of action.

Currently, we are trying to nail down a detailed description of the
driver constraint that will serve as a guideline for any proposal.  The
driver constraint will be a long list of conditionals going 'if we did
action set X, in Y years, change set Z will be possible.'  

The purpose of this post is not to take a position.  I'm just trying to
make sure that I'm following what's going on.  Maybe this post will
generate discussion that gives me(and hopefully some others) a clearer
understanding of what everyone is currently trying to do here.  Thanks.


Dan Jen


--
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