Continuation complexity: A callback hell for distributed systems

Abstract

Designing and validating large-scale distributed systems is still a complex issue. The asynchronous event-based nature of distributed communications makes these systems complex to implement, debug and test. In this article, we introduce the continuation complexity problem, that arises when synchronous invocations must be converted to asynchronous event code. This problem appears in most Actor libraries where communication is mainly asynchronous, and where a synchronous call to other Actor would block the current Actor, precluding the processing of incoming messages. We propose here a novel parallel invocation abstraction that considerably simplifies the continuation complexity problem for distributed actor systems requiring non-blocking synchronous invocations. Our parallel abstraction extends the message passing concurrency model to support concurrent interleavings of method executions within a single Actor. We present here two candidate solutions for implementing such parallel calls: one based on threads and locking, and other based on green threads and continuations. We evaluated the simplicity of our solution implementing a well known distributed algorithm like Chord (ring-based structured overlay). We compare our Actor implementation of Chord with three different simulators (PeerSim, PlanetSim, Macedon). This validation demonstrates that our approach is superior in simplicity (less LoC) and complexity (less McAbe complexity), envisaging its great potential for distributed systems scenarios.