Summary: Review of a Layered Naming Architecture for the Internet
March 10, 2008
The paper investigates the current name resolution methodology in the Internet and
argues that three layered name resolution architecture is required for mobility,
multihoming, and flat namespace for services and data. The main premise of their
argument is that data and services are named relative to the hosts on which they reside
and this strict binding of data and services to host restricts mobility of services. They
propose four design principles to overcome the above mentioned problem. The four
design principles are as follows:
· Names should bind protocols only to relevant aspects of the underlying
infrastructure and unwanted binding of protocol to other structures limits
flexibility and functionality.
· Persistent names should not impose arbitrary restriction on the elements to which
· A network entity should be able to direct its name resolution to all locations
(including itself) that run the same services.
· Senders and receivers should be able to loosely dictate the path of packets
sent/received by them by resolving to a sequence of identifiers instead of
resolving to a particular identifier (or location).