Practical Coded Sharding

Blockchain technology is finding applications not only in the cryptocurrency world but also in sectors like healthcare, IoT etc. However, before it can replace centralized counterparts like VISA or PayPal, its transaction throughput has to increase multiple orders of magnitude from its current throughput. Sharding is one solution for increasing the trans-action rate but it weakens the underlying security guarantees. Coded sharding uses information-theoretic approaches to scale blockchains without compromising security. Deploying coded sharding at webscale remains a challenge since it incurs bottlenecks like decoding very large Reed -Solomon codes of length of as much as 218. In this work, we propose a new method that leverages verifiable computing as an alternative to costly decoding strategies. We also simulate a simple balance checking application using our approach to check the feasibility of using verifiable computing with coded sharding. As its overheads are lower, we report transaction rates of 2806 tx/sec on commodity hardware. We also report the throughput of coded sharding based on verifiable computing by varying parameters such as the number of shards, users handled by each shard and complexity of verification used.