[arin-ppml] Routing Research Group is about to decide its scalable routing recommendation

[Robin Whittle]

Hi Bill,

You wrote, in part:

>    Efficient Internet routing is all about aggregation.  Combine
>    multiple downstream routes into a single CIDR prefix sent upstream
>    and all is well in the world.  Fail and the routing table grows, the
>    cost of routers rises and the general stability of the Internet falls.

Yes, assuming we keep the current interdomain routing system with its
BGP-based DFZ.  I support these remaining.  No-one has a practical
way of introducing something else.

Also, it is harder for the increasing number of end-user networks
which need multihoming and portability to get it.


>    Because of how TCP and the various UDP-based transport protocols
>    interact with the IP address, multihoming and mobility can defeat
>    aggregation.  

Because the only way of doing multihoming at present is to get your
own PI space and advertise it in the DFZ.

>    These protocols require the IP address to remain the
>    same throughout their operation.  Any time a host changes its
>    location within the network without also changing its IP address,
>    knowledge about that address becomes disaggregate with its neighbors.
>    The change must be propagated throughout the entire network.

Yes - there's no solid estimate of the number of DFZ routers, but
123k was the best estimate anyone could find in August 2007:

  http://www.ops.ietf.org/lists/rrg/2007/msg00253.html

Every one of those DFZ routers has to develop a best path for every
end-user network - and may have to do extra work whenever that
network advertises its prefix from a different point in the topology.


>    Routing researchers believe that if the host could readily change its
>    IP addresses to match its attachment to the network then the network
>    wouldn't have to change its routing to match the host's movement.
>    This would improve aggregation and reduce the frequency of routing
>    updates needed to keep the network operating.  They call this concept
>    "locator/identifier separation."

This is a host-based approach, such as with HIP.  Within the RRG and
the scalable routing field, "Locator/Identifier Separation" is often
used to refer to core-edge separation schemes (see my previous
message to Leo Bicknell), especially since one of them is called
(the) Locator Identifier Separation Protocol (LISP).  But LISP
doesn't really do this.  HIP is the real deal for Locator/Identifier
Separation.

So I agree with your use of the term.  I am just noting that I think
the term has been widely misused in recent years.

I also agree that if hosts did this, we could keep the existing
interdomain routing system and put all end-user networks on
inherently scalable PA space from their ISP.  Then, multihoming and
inbound TE involves switching a whole network from one ISP's PA
prefix to another.  Portability is portability of the logical address
space assigned to the end-user network, which is a different kind of
address space - in a different namespace - from the physical
connection addresses of the PA space.  (It would also be possible to
do it with locator and identifier addresses being of the same type:
sharing the same namespace - just different subsets of the one range
of numbers.)

However, as noted below, I have strong objections to all hosts being
required to do this.  So I support a core-edge separation approach
instead.


>    Locator/identifier separation's premise is simple: don't use the IP
>    address for both forwarding packets through the network and
>    associating those packets with their respective endpoints.  Instead,
>    separate this overloaded functionality into distinct elements within
>    each packet: locators used solely for forwarding packets and
>    identifiers used to associate those packets with specific hosts,
>    services and sessions.
> 
>    Practically speaking, this means we can either treat the IP address
>    as a host identifier and build an overlay to the routing system with
>    a new locator field somewhere in the packet or we can treat the IP
>    address as a locator and introduce new elements into the transport
>    protocols to figure out which packets belong to who.

In a theoretical sense, I agree this would be a good thing to do:
keep the network simple and make the hosts do more work.  This is
what made the Internet such a great thing compared to the phone network.

Every host gets a potentially unstable PA address, and can move
itself to any other PA address while retaining its application level
address and continuing host-to-host sessions.

However, this is at odds with my argument that it is wrong to burden
all hosts with such extra responsibilities.  I summarised these
objections in my previous message to Leo.  They are at:

  http://www.firstpr.com.au/ip/ivip/RRG-2009/host-responsibilities/

 - Robin