An implementation of complete , asynchronous , distributed garbage collectionFabrice

Most existing reference-based distributed object systems include some kind of acyclic garbage collection, but fail to provide acceptable collection of cyclic garbage. Those that do provide such GC currently suuer from one or more problems: synchronous operation, the need for expensive global consensus or termination algorithms, susceptibility to communication problems, or an algorithm that does not scale. We present a simple, complete, fault-tolerant, asynchronous extension to the (acyclic) cleanup protocol of the SSP Chains system. This extension is scalable, consumes few resources, and could easily be adapted to work in other reference-based distributed object systems|rendering them usable for very large-scale applications. 1 Introduction Automatic garbage collection is an important feature for modern high-level languages. Although there is a lot of accumulated experience in local garbage collection, distributed programming still lacks eeective cyclic garbage collection. A local garbage collector should be correct and complete. A distributed garbage collector should also be asynchronous (other spaces continue to work during a local garbage collection in one space), fault-tolerant (it works even with unreliable communications and space crashes), and scalable (since networks are connecting larger numbers of computers over increasing distances). Previously published distributed garbage collection algorithms fail in one or more of these requirements. In this paper we present a distributed garbage collector for distributed languages that provides all three of these desired properties. Moreover, the algorithm is simple to implement and consumes very few resources. The algorithm described in this paper was developed as part of a reference-based distributed object system for Objective-CAML (a dialect of ML with object-oriented ex