Memory space representation for heterogeneous network process migration

A major difficulty of heterogeneous process migration is how to collect advanced dynamic data-structures, transform them into machine independent form, and restore them appropriately in a different hardware and software environment. In this study we introduce a data model, the Memory Space Representation (MSR) model, to recognize complex data structures in program address spaces. Supporting mechanisms of the MSR model are also developed for collecting program data structures and restoring them in a heterogeneous environment. The MSR design has been implemented under a prototype heterogeneous process migration environment. Pointer-intensive programs with function and recursion calls are tested. Experimental results confirm that the newly proposed design is feasible and effective for heterogeneous network process migration