DECK: A new model for a distributed executive kernel integrating communication and multithreading for support of distributed object oriented application with fault tolerance support

DECK (Distributed Executive Communication Kernel) is a communication layer that provides support for multithreading and fault tolerance support. The approach retained in DECK is close to other distributed communication kernels like PM2, Athapascan, Nexus, TPVM or Chant in its way to integrate communication and multithreading to efficiently overlap communication by computation and provide low latency remote thread creation mechanisms. However, DECK differs from these communication kernels from the services offered and its modular architecture. The main goal of DECK is to implement a new model for the design of distributed executive kernel to efficiently use the new underlying hardware architectures (SMP architectures and fast communication adapters like Myrinet or memory oriented adapter like SCI) and provide a portable layer that abstract the problems linked with the integration of communication and multithreading while offering support for heterogeneity. A great lack in the current implementation of communication libraries or distributed executive kernel is the support for basic services at the thread level and support for fault tolerance support. Indeed, communication library like PVM or MPI are often used as communication layer to ensure portability and take benefits of specific implementation to ensure a good efficiency on specific architectures however the support for fault tolerance support, multithreading, scalability and interoperability are usually not offered. In the case of DECK, we propose a model where a distributed application can dynamically instantiate clusters of processes among an heterogeneous network of computers or parallel machines and this using multiple communication protocols or communication interfaces to ensure good performances regarding the underlying hardware architecture. The programming model proposed offer both classic synchronous and asynchronous remote service calls for thread creation and message passing for synchronization and data exchange. These basic functionalities, that form the low level communication and execution layer of DECK, are enforced by a service layer that propose the basic fault tolerant services like naming and group services or data management services for the marshaling and un-marshalling of complex data structures. The layered and modular approach followed by DECK enable many other extensions while keeping a high degree of portability and efficiency. Sistemas Distribuidos - Redes Concurrencia Red de Universidades con Carreras en Informática (RedUNCI)