Tutorial 1: Synchronization is coming back but is it the same?
- Michel Raynal, IRISA/Univ. Rennes 1, France
Informally, "wait-free" means that the progress of a process depends only on itself.This notion is more and more pervasive in a lot of problems that basically rely (in one way or another) on the definition and the use of concurrent objects in presence of failures. This tutorial will visit wait-free computing: its underlying concepts and its basic mechanisms. To that end, the lecture will also visit fundamental problems of asynchronous computing in presence of failures such as renaming, set agreement, collect, snapshot, etc. It will also present fundamental notions related to the implementation of concurrent objects, such as t-resilience and graceful degradation.
The literature on this topic is mostly technical and appears mainly in theoretically inclined journals and conferences. The aim of this tutorial is to offer an introductory survey to the new synchronization concepts that have been introduced in the recent past. The tutorial is destined to the people who are not familiar with these concepts and want to quickly understand their aim, their basic principles, their power and limitations. The tutorial will adopt an algorithmic approach to explain these new concepts. From a practical point of view, the advent of multicore architecture makes this topic central for researchers and engineers whose main interests lie in distributed fault- tolerance and dependability for shared memory systems. Moreover, whatever the problem they have to solve, one aim of the tutorial is to enlarge the knowledge and the background of researchers and engineers whose main interest is dependability.