Software pattern design for cluster computing

While using a cluster of computers in a network environment several issues need to be considered: (i) choice of models for software architectural pattern, (ii) program serializability, fault tolerance and recovery, (iii) choice of programming paradigms, (iv) factors affecting scalability in performance-such as transmission and computation times, (v) dangers of cumulative errors, (vi) compatibility and interoperability. The paper describes some basic models for cluster computing. Also it considers the issues of global serializability and scalability and provides some useful insights on the evolution of new practical software tools.