Rapid Blockchain Scaling with Efficient Transaction Assignment

Abstract

The abrupt development of cryptocurrencies and blockchain technology has pointed to the importance of deploying large scale, highly robust Byzantine fault-tolerant schemes to handle critical distributed failures in system networks. Although traditional wisdom is to build systems by following synchronous protocols such as PBFT consensus, these protocols strongly rely on network time limitations and only guarantee liveness when the nodes behave arbitrary. This paper presents the design and implementation of Adrastus, a scalable blockchain system. The main contributions of Adrastus are the use of a consistent hashing mechanism which solves load balancing problems and the efficient assignment of transactions on parallel regions of single-chain consensus systems, referred to as zones, without introducing unnecessary overhead. We claim that the Adrastus blockchain system scales linearly without compromising system security. We present theoretical analysis, discuss our solution, and examine the conditions to meet both safety and liveness of our fault-tolerant system.