BATRUN DPS (Batch after twilight running-distributed processing system): multi-cell, fault-tolerant, parallel, batch processing for Monte Carlos simulations

This paper discusses the design of the BATRUN Distributed Processing System (DPS). In contrast to a dedicated cluster of workstations, the scheduling in BATRUN DPS must ensure that only the idle cycles are used for distributed computing and the local users, when they are operating, have full control of their machines. BATRUN DPS has several unique features: group-based scheduling policy to ensure execution priority based on ownership of machines, and multi-cell distributed design to eliminate a single point failure as well as to ensure scalibility. The implementation of the system is based on multi-threading and remote procedure call mechanisms. (author)