Re: SIP Addressing Limitations

["Paul Francis (formerly Paul Tsuchiya" <[e-mail]>]

>  
>  
>  > hosts), the SIP address could handle 10^12 hosts if it acheived
>  > about 2% efficiency per level of hierarchy.  By 2% efficiency
>  > I mean that 2% of the possible address assignments (per level)
>  > are actually assigned.
>  >
>  Then you will be glad to have read:
>  
>        While at first glance this may appear to be 89% efficiency in
>        allocation, it is noted below that the two largest countries are
>        assigned only half their proper allocation.  The actual efficiency
>        is closer to 60%, ....

I wondered about this.  What do you mean by 60% efficiency?

I think it is impossible to talk about efficiency until you
start really loading up an address space.  For instance, if
China and the existing third world stayed technologically 
poor for the next 20 years, while first world countries starting
using some new technology that resulted in enormous numbers of
address assignments, then your efficiency measure for those
portions of the address assigned to third world countries
would have low efficiency......

>  
>  Paul, in my view it's time for you to put up or shut up.  Show me this
>  fine allocation plan of yours with:
>  
>   - a guaranteed set of constraints.
>   - a provable maximal routing table size.
>   - aggregates so that an international amateur radio operator sees a
>     very small table (< 256 entries, preferably < 10).
>   - scales on an inter-planetary basis.
>  
>  And is:
>  
>   - easy to administer.
>   - amenable of politics.
>   - supports metro, provider, and end-point allocation simultanously.
>   - has a firm rationale for each division.
>   - supports efficient distributed access.
>   - aggregates better as the network grows.
>  

I can't claim to have a plan that has all of these attributes, and
I would argue that you don't either.

I like provider-based addressing because it doesn't depend on
the topology of providers, because I think it scales well over
the near term, and because the provider numbers can be used as
a hook for provider selection.

I like Pip addressing because you don't have to go through
the exercise you went for for SIP addresses--each level of
the hierarchy carried explicitly in the packet so you don't
have to worry about running out of space in one part of the
hierarchy or another.

I think it is virtually impossible to avoid widespread and
periodic number reassignments in any scheme, and I don't even
try.  Rather, Pip strives to make number re-assignment as
automatic and painless as possible.  The SIP people are going
to provide this for SIP as far as I know.  It is mainly because
of this that I don't object to much to the SIP addressing
scheme (I don't mean your particular allocation--I mean 64-bits
in general).  That is, provided they have automatic prefix
assignment, it will be possible to change the addressing scheme
over time.

>  Otherwise, give me *contructive* criticism for my plan.
>

For what your plan purports to do, it seems to do a good job.
You have set out to come up with a numbering plan that allows
for onion-skins of aggregation based on geographic/political
closenss.  You seem to have done a nice job of it.

I just generally disagree with your premise--that geographic/political
closeness is of primary interest.  

As this discussion has already been had, I will shut-up.....

PX