On Decoupling Concurrency Control from Recovery in Database Repositories
We report on initial research on the concurrency control issue of compiled database applications. Such applications have a repository style of architecture in which a collection of software modules operate on a common database in terms of a set of predefined transaction types, an architectural view that is useful for the deployment of database technology to embedded control programs. We focus on decoupling concurrency control from any functionality relating to recovery. Such decoupling facilitates the compile-time query optimization. <br /><br /> Because it is the possibility of transaction aborts for deadlock resolution that makes the recovery subsystem necessary, we choose the deadlock-free tree locking (TL) scheme for our purpose. With the knowledge of transaction workload, efficacious lock trees for runtime control can be determined at compile-time. We have designed compile-time algorithms to generate the lock tree and other relevant data structures, and runtime locking/unlocking algorithms based on such structures. We have further explored how to insert the lock steps into the transaction types at compile time. <br /><br /> To conduct our simulation experiments to evaluate the performance of TL, we have designed two workloads. The first one is from the OLTP benchmark TPC-C. The second is from the open-source operating system MINIX. Our experimental results show TL produces better throughput than the traditional two-phase locking (2PL) when the transactions are write-only; and for main-memory data, TL performs comparably to 2PL even in workloads with many reads.
Cite this version of the work
Heng Yu (2005). On Decoupling Concurrency Control from Recovery in Database Repositories. UWSpace. http://hdl.handle.net/10012/1084