A High-Level Checkpoint Thread Based Mechanism for Fault Tolerant Parallel Text Matching

摘要

A thread based, high level, checkpointing approach for data monitoring and recovery is investigated herein. This mechanism is applied on the parallel computation of the exact pattern matching problem over a cluster of workstations, utilizing a straightforward Brute Force algorithm. The resulting implementation is evaluated both theoretically and experimentally and the results are compared to a non fault-tolerant approach, so as to justify added functionality in contrast to performance.