Ensuring the Safety of Distributed Garbage Collection in DMOS

An orthogonally persistent system maintains data beyond any program execution. The principle of persistence identification implies that persistence by reachability be used to determine which objects are to be maintained in the store. A particular execution may cause objects to become unreachable—garbage. Any efficient implementation of a persistent system must incorporate a garbage collector to remove these objects. The availability of commodity machines and sophisticated networking technology makes it possible to construct low cost distributed memory machines. One way to harness the scalability of these machines is with a single image object store. However, the resulting distribution of data increases the system complexity. A distributed garbage collector must deal with safely updating meta-data across distributed processors as well as detecting cyclic garbage that exists on multiple processors. DMOS [Hudson et al. 1997] is a distributed garbage collector that is derived from the MOS and PMOS garbage collectors. DMOS is one of the few collectors that is distributed, asynchronous, non-disruptive and complete. DMOS involves several interacting protocols, making it difficult to understand and implement. Close examination of these protocols reveal deficiencies in the information propagated, and a race condition between the train isolation detection protocol and the reference information protocol. This paper addresses these issues, augments the information propagated and modifies the termination algorithm to ensure that the collector behavior is safe and correct. These corrections and additions enable the DMOS algorithm to be realized. As a result, it is possible to approach the goal of implementing a complete distributed persistent object store.