Concurrent linearizable access to shared objects can be prohibitively expensive in a high contention workload. Many applications apply ad-hoc techniques to eliminate the need of synchronous atomic updates, which may result in non-linearizable implementations. We propose a new programming model which leverages such patterns for concurrent access to objects in a shared memory system. In this model, each thread maintains different views on the shared object - a thread-local view and a global view. As the thread-local view is not shared, it can be updated without incurring synchronization costs. These local updates become visible to other threads only after the thread-local view is merged with the global view. This enables better performance at the expense of linearizability. We show that it is possible to maintain thread-local views and to perform merge efficiently for several data types and evaluate their performance and scalability compared to linearizable implementations. Further, we discuss the consistency semantics of the data types and the associated programming model.
Download