Actor Garbage Collection in Distributed Systems using Graph Transformation

TL;DR

This paper proposes a graph transformation approach for actor garbage collection in distributed systems, enabling the use of passive reference graph algorithms by transforming active actor references into passive graphs, addressing transformation costs and optimization.

Contribution

It introduces a novel graph transformation method to facilitate garbage collection for actors in distributed systems, overcoming limitations of traditional algorithms.

Findings

Transformation reduces garbage collection overhead

Graph-based approach improves collection accuracy

Optimization techniques mitigate transformation costs

Abstract

A lot of research work has been done in the area of Garbage collection for both uniprocessor and distributed systems. Actors are associated with activity (thread) and hence usual garbage collection algorithms cannot be applied for them. Hence a separate algorithm should be used to collect them. If we transform the active reference graph into a graph which captures all the features of actors and looks like passive reference graph then any passive reference graph algorithm can be applied for it. But the cost of transformation and optimization are the core issues. An attempt has been made to walk through these issues.