Data Redistribution and Remote Method Invocation in Parallel Component Architectures

Abstract

With the increasing availability of high-performance massively parallel computer systems, the prevalence of sophisticated scientific simulation has grown rapidly. The complexity of the scientific models being simulated has also evolved, leading to a variety of coupled multi-physics simulation codes. Such cooperating parallel programs require fundamentally new interaction capabilities, to efficiently exchange parallel data structures and collectively invoke methods across programs. So-called "M×N" research, as part of the Common Component Architecture (CCA) effort, addresses these special and challenging needs, to provide generalized interfaces and tools that support flexible parallel data redistribution and parallel remote method invocation. Using this technology, distinct simulation codes with disparate distributed data decompositions can work together to achieve greater scientific discoveries.

Collective Interfaces for Distributed Components

“…We plan to apply collective interfaces for facilitating the coupling of parallel codes (also referred to as the "MxN problem" [6]). We intend to develop more complex and varied applications, in particular the NAS Grid benchmarks [11], for which we can take advantage of both the hierarchical nature of the Fractal component model and the collective interfaces.…”

“…ICENI [16] proposes a range of dispatch modes, including scatter, gather, dispatch and reduce, but they are only applicable to arrays and require XML data. Several CCA frameworks [6] propose collective communications in order to provide efficient and modular code coupling, usually using intermediate components, and also through group method invocations. GridCCM [20] extends the Corba Component Model in order to efficiently couple parallel components.…”

TGrid - Grid runtime support for hierarchically structured task-parallel programs

“…As for TGrid, the M ×N problem (data redistribution) plays an important role for CCA components. The requirements for creating a data redistribution framework for CCA components is addressed in [4].…”

Design and Evaluation of a Parallel Data Redistribution Component for TGrid

“…An ongoing research project is carried out by the CCA (common component architecture) forum. The MxN working group of the CCA forum is working on the definition and implementation of interfaces to transfer data elements between parallel components running with different numbers of processes in each parallel component [8]. The framework Seine [9] is a geometry-based interaction model which is encapsulated as a CCA compliant component within the Ccaffeine CCA framework.…”