Authored by:
Yiding Hua, ETH Zurich
Deepal Tennakoon, University of Sydney
Vincent Gramoli, University of Sydney, CTO Redbelly Network
Decentralisation promises to remedy the drawbacks of the web by executing decentralised applications (DApps) on blockchains. Unfortunately, modern blockchains cannot support realistic web application workloads mainly due to congestion.
We introduce the Smart Redbelly Blockchain (SRBB), a provably correct permissionless blockchain that reduces congestion by (1) avoiding redundant propagation and validations of transactions with Transaction Validation and Propagation Reduction (TVPR) and (2) mitigating the propagation of invalid transactions within blocks by Byzantine nodes with a dedicated Reward Penalty Mechanism (RPM). Our comparison of SRBB against Algorand, Avalanche, Diem, Ethereum, Quorum, and Solana, using the DIABLO benchmark suite, indicates that SRBB outperforms all these blockchains under real application workloads. Moreover, SRBB is the only blockchain to successfully execute real workloads of NASDAQ and Uber on a DApp without losing transactions. To demonstrate that TVPR and RPM are the causes of the improved performance, we compare SRBB with its naive baseline, which does not contain TVPR and RPM. Our results show that TVPR increases the throughput by 55× and divides the latency by 3.5, while RPM increases the throughput by 7% under flooding attacks. Finally, TVPR helps reduce transaction losses in the normal scenario while RPM goes further and mitigates transaction losses under flooding attacks.
This paper was co-authored by our CTO and Founder Professor Vincent Gramoli.
Professor Vincent Gramoli is a full professor at the University of Sydney. He is a researcher in the field of distributed systems and algorithms, with a focus on the design and analysis of distributed systems and algorithms for shared memory and data-centric systems, including distributed hash tables, distributed shared memory and transactional memory. He has published numerous papers in top-tier conferences and journals in the field and has received several awards for his research. He is also currently serving as the Head of Concurrent Systems Research Group at the University of Sydney.