Analyzing Soft and Hard Partitions of Global-Scale Blockchain Systems

Abstract

Partitioning attacks on blockchain systems are a serious threat, with the potential to cause significant harm on the individual level and to the system as a whole. Deliberate partitions can be used by attackers to defraud merchants using crypto-currencies and enrich themselves, while natural disasters and wars could disrupt blockchain systems for months, potentially destroying them entirely. Unfortunately, the exact effects of these partitions at large scale are unknown and difficult to model, making it challenging to implement preventative measures or plan for partition recovery. In this work, we examine a variety of real-world global scale partitioning events to develop a framework for categorizing and analyzing different partition types. We use this framework, along with a new metric introduced to finely measure the global coherence of a given blockchain, to quantify the impacts of soft and hard network partitions at varying scales. Our goal is to lay the groundwork for the introduction of new blockchain architectures to minimize the harm caused during partitions and facilitate rapid and uncontroversial recovery from them.