CoVer-ability: Consistent Versioning for Concurrent Objects

TL;DR

This paper introduces coverability, a new property for concurrent objects that ensures consistent evolution, and explores its levels, implementations, and equivalences, advancing distributed object guarantees.

Contribution

It defines coverability with strong and weak levels, analyzes their properties, and demonstrates their applications and limitations in distributed object implementations.

Findings

Weak coverability cannot be implemented using ranked-registers.

Weak coverability enables implementation of basic read-modify-write objects.

Strong coverability is equivalent in power to consensus.

Abstract

An object type characterizes the domain space and the operations that can be invoked on an object of that type. In this paper we introduce a new property for concurrent objects, we call coverability, that aims to provide precise guarantees on the consistent evolution of an object. This new property is suitable for a variety of distributed objects including concurrent file objects that demand operations to manipulate the latest version of the object. We propose two levels of coverability: (i) strong coverability and (ii) weak coverability. Strong coverability requires that only a single operation can modify the latest version of the object, i.e. "covers" the latest version with a new version, imposing a total order on object modifications. Weak coverability relaxes the strong requirements of strong coverability and allows multiple operations to modify the same version of an object, where each modification leads to a different version. Weak coverability preserves consistent evolution of the object, by demanding any subsequent operation to only modify one of the newly introduced versions. Coverability combined with atomic guarantees yield to coverable atomic read/write registers. We also show that strongly coverable atomic registers are equivalent in power to consensus. Thus, we focus on weakly coverable registers, and we demonstrate their importance by showing that they cannot be implemented using similar types of registers, like ranked-registers. Furthermore we show that weakly coverable registers may be used to implement basic (weak) read-modify-write and file objects. Finally, we implement weakly coverable registers by modifying an existing MWMR atomic register implementation.