Scheduling Closed-Nested Transactions in Distributed Transactional Memory

Distributed software transactional memory (D-STM) is an emerging, alternative concurrency control model for distributed systems that promises to alleviate the difficulties of lock-based distributed synchronization -- e.g., distributed deadlocks, live locks, and lock convoying. We consider Herlihy and Sun´s dataflow D-STM model, where objects are migrated to invoking transactions, and the emph{closed nesting} model of managing inner (distributed) transactions. We present a transactional scheduler called, reactive transactional scheduler (or RTS) to boost the throughput of closed-nested transactions. RTS determines whether a conflicting parent transaction must be aborted or enqueued according to the level of contention. If a transaction is enqueued, its nested inner transactions do not have to retrieve objects again, resulting in reduced communication delays. Our implementation of RTS in the HyFlow D-STM framework and experimental evaluations reveal that RTS improves throughput over D-STM without RTS, by as much as 88%.