Transactional Memory


    The spread of multiprocessor architecture will have a pervasive effect on how software is developed. Multiprocessors make computing more effective by exploiting parallelism. Harnessing the architecture of concurrent programming and multiple thread operation sequencing realizes high performance. Increasing the number of processors is not a panacea for high performance, but combining this with advanced concurrency control leverages performance. As a traditional synchronization scheme, lock-based concurrency control suffers from programmability, scalability, and compensability challenges. Transactional memory (TM) has been motivated by these difficulties. The challenges of lock-based concurrency control are exacerbated in distributed systems due to the additional complexity of multicomputer concurrency. Distributed TM (or DTM) has been similarly motivated as an alternative to distributed lock-based concurrency control. DTM is an emerging, alternative concurrency control model that promises to alleviate the difficulties of lock-based distributed synchronization. A complementary approach for handling conflicts is through a transactional scheduler, which orders transactional requests to avoid or minimize conflicts. Dr. Kim proposed a suite of transactional schedulers in the full and partial replicated models of distributed systems.