Flask : an architecture supporting concurrent distributed persistent applications

Abstract

Distributed application systems have become a popular and provenly viable computing paradigm. There are a number of reasons for this such as: the geographical dispersal of information; the improved reliability of multiple computer systems; and the possibility of concurrent execution of applications. As yet no single model of distribution has been pervasive and since the impact of failure semantics varies with the software architecture of applications, it is unlikely that one model will ever dominate. It is difficult to assess or even to compare the attributes of different models especially when run over the same data. This is often made more difficult in that most implementations of distributed models are closed systems with built-in protocols, failure reporting and concurrency control. The Flask architecture, presented here, takes the approach of providing a layered architecture which has the flexibility to support different models of distribution that can run over the same data. To demonstrate the feasibility of Flask an example distributed application is described using the architecture.