The Road to El Diablo: Towards Secure High Performance Blockchains

Abstract

It is no exaggeration to say that Bitcoin constituted the reboot of research efforts in distributed systems and Byzantine fault tolerance. However, rising popularity resulted in a flurry of proposals, each aiming to address problems in the initial specification by Bitcoin, or, introduce new functionality. This added considerable complexity to the blockchain ecosystem, amplified by the absence of detail in accompanying documentation, whitepapers or blog posts. Although moving towards technological maturity, the blockchain presents new environments for distributed systems with implicit assumptions, with developers often overlooking critical details that lead to vulnerabilities and mistaken guarantees in practice. This dissertation presents fundamental contributions towards secure, high performance blockchains.

Firstly, we evaluate the impact of misrepresented assumptions, identifying anomalous behaviour as a result of chain reorganisation; indicating weaknesses of probabilistic blockchain consensus. We develop and propose the Balance Attack, a novel attack that utilises communication across multiple subgroups of nodes to double-spend, emphasising the importance of critical assumptions and guarantees, such as synchrony and committing transactions. This provides the foundations for the Red Belly Blockchain, a secure, high performance blockchain which we present to mitigate risks posed by anomalies and attacks such as the Balance Attack. Experimental evaluation shows that the Red Belly Blockchain improves upon performance offered by other Byzantine fault tolerant blockchains while scaling to hundreds of consensus nodes. Finally, we developed DIABLO, a modular, distributed benchmark framework capable of facilitating fair and accurate comparisons of blockchains through dynamic, real world application workloads. This provides integral performance validation of our Red Belly design and paves the way for future analysis and further comparison of new blockchains and workloads.