Checkpointing BSP parallel applications on the InteGrade Grid middleware

摘要

InteGrade is a Grid computing system aimed at commodity workstations such as household PCs, corporate employee workstations, and PCs in shared university laboratories. InteGrade uses the idle computing power of these machines to perform useful computation. The goal is to allow organizations to use their existing computing infrastructure to perform useful computation, without requiring the purchase of additional hardware. Running scientific applications over shared workstations requires a sophisticated software infrastructure. Users who share the idle portion of their resources should have their quality of service preserved. If an application process is running on a previously idle machine whose resources are requested back by its owner, the process should stop its execution immediately to preserve the local user's quality of service. In the case of a non-trivial parallel application consisting of many processes, stopping a single process usually requires the reinitialization of the whole application. Mechanisms such as checkpoint-based rollback recovery can be implemented in order to solve this kind of problem.