Maximizing concurrency in predicate locking

The paper investigates a central aspect of predicate locking, studying u nd er which conditions a schedular can achieve the highest concurrency. A scheduling strategy allowing the maximum concurrency is derived, showing that the feability of this strategy depends on the tractability of the langu agc for expressing predicate locks. Then a predicate language allowing all efficient implemeutation of the optimal strategy is identified, and the implementation is exemplified. Fillally, the range of applicability of the optimal strategy is outilines by showing that significant ex t cnsions to the expressive power of the previously introduced predicate languages causes all irreparable loss efficency.