Review: IPNL: A NAT-Extended Internet Architecture

From: Di Niu <dniu_at_eecg.toronto.edu>
Date: Thu, 30 Nov 2006 11:25:21 -0500

Review: IPNL: A NAT-Extended Internet Architecture

Reviewer: Di Niu

This paper presents and analyzes IP Next Layer (IPNL), a NAT-extended
Internet protocol architecture designed to scalably sovle the address
depletion problem of IPV4. Because of the spread of Network Address
Translator (NAT), the Internet suffers more and more from the loss of
end-to-end addressability. The main benefits of NAT are that it
expands the IPv4 address space and that it isolates a site's address
space from the global address space. Because of this address
isolation, a NAT'ed site can be attahced to multiple ISPs without
having the site's address prefix advertised across the default-free
routing zone of the Internet. And thus, NAT is argued to be a key
technology responsible for what limited scalability of the Internet.
Two primary negative aspects of NAT are that it inhibits the
introduction of certain kinds of peer-to-peer applications and that
it complicates scalable network operation and new protocol and
application design.

To address these problems, the paper proposes IPNL, which is an
extension to NAT. The major attributes of IPNL are as follows: First,
it is a NAT-extended architecture, which means that it maximizes
reuse of the existing IPv4 infrastructure, primarily by adding a new
layer above IPv4 that is routed by NAT boxes. Second, it utilizes
Fully Qualified Domain Names (FQDNs) as an end-to-end host identifier
in packets. Third, it 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 its low order
part. Fourth, it completely isolates site addressing from blobal
addressing.

This is a pretty-good paper which at least contributes to the
networking community by pointing out the important problems caused by
NAT. Although IPNL certainly does not servers as a terminator in the
long run of solving these problems, it could be a good start point,
as it has a number of interesting characteristics, such as the
various mechanisms for site isolation and scalable multihoming.
However, the implementation of IPNL will remain to be a hard task.
Received on Thu Nov 30 2006 - 11:26:36 EST

This archive was generated by hypermail 2.2.0 : Thu Nov 30 2006 - 11:58:36 EST