Mobile Calculi based on Domains


Building Reliable Distributed Infrastructures Revisited: a Case Study

M. Lacoste

Abstract

Building reliable distributed infrastructures still remains a challenge. Two separate and slowly diverging visions for solutions have emerged: the first relies upon an implementation-based approach of today's now mature middleware object technologies, often without much concern for the underlying theoretical foundations. The second focuses on elaborate theoretical models of distributed computation, to capture various aspects of communication, mobility, or security, but has produced few implementations. We attempt to reconcile on a case study those two opposite views of reliable distributed computing using a simple methodology: define both a carefully designed distributed computing model with a well-defined semantics and a middleware implementation that strictly conforms to the model. We demonstrate the feasibility and the soundness of such a formal approach by describing the refinement steps in the formalization, from the model design to the implementation, using as an example a simple distributed programming model named DCP. In DCP, remote interaction is primitive and the distribution of resources is explicit while keeping network communications transparent to the programmer. The DCP implementation is formalized by a distributed abstract machine, and seamlessly integrated with a middleware representative of current distributed technology -- the Jonathan Object Request Broker -- to support network transparent communications.

@InProceedings\{lacoste:building-reliable-infrastructures,
  author = \{M. Lacoste},
  title = \{Building Reliable Distributed Infrastructures Revisited: a Case Study},
  booktitle = \{International DOA Workshop on Foundations of Middleware Technologies (WFoMT'02)},
  year = \{2002}, 
  url = \{http://mikado.di.fc.ul.pt/repository/lacoste_building-reliable-infrastructures.pdf}
}


About this site. Last modified: Wed Sep 20 22:10:06 CEST 2017