Review: IPNL: A NAT-Extended Internt Architecture

NAT has become significanly popular in the Internet and resulted in loss
of end-to-end addressability. The paper presents a NAT-extended Internet
protocol called IPNL (IP Next Layer) to scalably solve the address
depletion problem of IPV4. The authors set out to explore whether a
NAT-extension would make a uitable Internet architecture. This involves
two question:
1. What constitutes a suitable Internet architecture, which they define
as one that preserves the original charactersitics of IPV4 while solving
its scalability and address depletion problems. Furthermore, the state
the charactersitics of IPV4 as 1)all hosts are identified by long-lived,
globally routable address 2)routers are stateless 3)address prefix is
independent of network attachment location and 4)packets cannot be
easily hijacked by rogue or misconfigured hosts that are not on the
physical path of the packets.
2. What constitues an extension of NAT, which is defined as one that
wokrd by modifying only hosts and NAT boxes.
The authors highlight the major attributes of IPNL:
1. NAT-extended architecture: maximizes reuse of existing IPV4
infrastructure by adding new layer below the transport layer (TCP/UDP)
and above the IP layer.
2. Utilizes Fully Qualifies Domain Names (FQDNs) as an end-to-end host
identifier.
3. Extends the IP address space such that the globally unique IP address
space forms the high order part of the IPNL address and the private IP
address space forms the lower order part.
4. Completely isolates site addressing from global addressing.
The paper emphasizes that the IPNL topology is the same as today's
Internet topology: privately-addressed realms connected to the
globally-addressed Internet and sometimes to each other through NAT
boxes. IPNL uses extended IP addresses consisting of a global IP, realm,
and local IP. The protocol's nl-routers route based on both FQDNs
(static long-term addresses) as well as IPNL addresses (short loved and
each host may have more than one which are bootstrapped and maintained
by the FQDN). Robustness is provided by additional path discovery and
in-band trace (allows hosts to quickly detect failure of an nl-router in
its realm or in a destination host's frontdoor).
The paper is well written and provides quite a detailed description of a
new architecture. However, towards the end of the paper, while talking
about next steps, the authors seem to start doubting their own design
and its deployability. They started off pushing the idea by mainly
saying it was deployable as compared to IPV6 but then pull back towards
the end. I think this leaves the reader doubting the whole architecture too.
Received on Thu Nov 30 2006 - 17:09:48 EST