[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[RRG] Geographic aggregation-based routing is at odds with reality



I think that geographical-based addressing arrangements, such as
those proposed in the recent thread by Heiner Hummel and Iljitsch
ban Beijnum) are not worth considering further because:

   To the extent that routing scalability depends on geographically
   determined assignment of address space, this is completely
   incompatible with several fundamental needs of providers and
   end-user networks:

     1 - That organisations who have been assigned address space
         should be free to use it at various sites, and these
         organisations are frequently global.

           (Otherwise, each branch of an organisation - and
            there could be hundreds at the granularity required
            by the geographic aggregation system - will want to
            get a large slab of address space, to cope with the
            potential for future expansion.)

     2 - Since the scalability which geographic aggregation
         supposedly must depend on routers forwarding packets
         in part or in whole according to their destination
         (source too??) address, this is incompatible with
         the need of organisations to have packets flow along
         paths which are determined by their business relationships.

           (It is assumed that the Internet's routing and addressing
           system should not require any organisations to have a
           business relationship or handle each other's packets
           simply because they are in some kind of geographic
           proximity.)

        Bill's challenge to Heiner illustrates point 2 nicely:

          http://psg.com/lists/rrg/2008/msg01815.html
          http://psg.com/lists/rrg/2008/msg01829.html


    3 - Organisations need to choose who they connect their networks
        to according to various criteria which are at odds with
        geographical aggregation, including being free to create
        links to distant networks.  Scenarios include:

          a - Redundant paths to cope with (geographically) nearby
              failures and points of congestion.

          b - Similarly, paths (such as by a fibre link, not
              tunnelling through the Net) which enable packets
              to travel whilst meeting security and policy needs.

                (For instance, for security - not through any
                given country or company.  Also, to meet local
                Internet censorship, anti-terrorism etc. laws,
                it may be necessary to make links which avoid
                certain countries.  Encryption is not a proper
                solution, and security can be damaged just
                by analysing traffic patterns, even if the
                contents cannot be deciphered.)

          c - Efficient traffic handling within global private
              networks which nonetheless use public address space.

Having the scalability of the Internet's routing system depend on
assigning addresses according to geographical location - implicitly
with forwarding of packets being dependent upon those addresses - is
completely incompatible with the business, policy, security and
efficiency requirements of the great majority of providers and
end-user networks.

Geographical aggregation is the sort of thing which looks good on
paper, but will never be acceptable in the real world.

  - 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