EAGER: Foundations of Data-Centric Concurrency Control

Purdue University, West Lafayette IN

Investigators

Abstract

This project investigates the foundational aspects of an alternative approach to the specification, implementation and management of concurrent activities which is referred to as data-centric synchronization. Traditional approaches take an operational view of concurrency control, they burden the programmer with the need to identify sets of control flow paths in their program which must not interfere. This has been shown to be difficult to get right and, more often than not, to inhibit scalability to multicore systems as the code is over-synchronized. This EAGER project explores the foundations of data-centric synchronization with an emphasis on establishing a type theoretic foundation based on the notion of ownership types. Ownership types give programmers a simple notation for describing the shape and extent of heap-based data structures. Thus an ownership type system can be seen as the basic mechanism for specifying the groups of data items that must be manipulated synchronously.