A Dynamic Recursive Unified Internet Design (DRUID)
Joe Touch, Ilia Baldine, Rudra Dutta, Gregory G. Finn, Bryan Ford,
Scott Jordan, Dan Massey, Abraham Matta, Christos Papadopoulos,
Peter Reiher, George Rouskas
Computer Networks
December 24, 2010
Abstract:
The Dynamic Recursive Unified Internet Design (DRUID) is a future Internet
design that unifies overlay networks with conventional layered network
architectures. DRUID is based on the fundamental concept of recursion, enabling
a simple and direct network architecture that unifies the data, control,
management, and security aspects of the current Internet, leading to a more
trustworthy network. DRUID's architecture is based on a single recursive block
that can adapt to support a variety of communication functions, including
parameterized mechanisms for hard/soft state, flow and congestion control,
sequence control, fragmentation and reassembly, compression, encryption, and
error recovery. This recursion is guided by the structure of a graph of
translation tables that help compartmentalize the scope of various functions
and identifier spaces, while relating these spaces for resource discovery,
resolution, and routing. The graph also organizes persistent state that
coordinates behavior between individual data events (e.g., coordinating packets
as a connection), among different associations (e.g., between connections), as
well as helping optimize the recursive discovery process through caching, and
supporting prefetching and distributed pre-coordination. This paper describes
the DRUID architecture composed of these three parts (recursive block,
translation tables, persistent state), and highlights its goals and benefits,
including unifying the data, control, management, and security planes currently
considered orthogonal aspects of network architecture.
Journal Preprint:
PDF