Self-organizing systems offer an alternative to traditional distributed systems in those cases where central coordination or central databases, pre-existing fixed network infrastructure, and centralized services are precluded.

In self-organized systems, the peer nodes act as both clients and servers, and none of them has a global view of the system. The need to provide efficient data availability is usually solved by replicating the data across the network. Unfortunately, this approach wastes resources. To address this problem, the author of this paper, who is also interested in modeling data access for mobile ad hoc networks, presents the details of how he created a distributed system annex prototype that allowed him to test his particular approach.

The author discusses the conceptual foundations of the system, stressing real-time access and reliable data access in constantly evolving environments. He also discusses the issues raised by the concepts, and the ways in which the issues were addressed. He then describes the architecture and the design methodology, and how it focuses on the distributed access to mobile data involved in the computation of a multi-point relay, emphasizing that nodes can leave the network without warning, thus making it necessary for all neighboring nodes to quickly select other multi-point relays. A brief description of the various kinds of software components used in the model is also provided.

The paper is concise and well written, with an excellent conclusion.