Demystifying Blockchain Scalability: Sibling Chains with Minimal Interleaving

Blockchain provides alluring infrastructure for distributed ledgers supporting anonymous online payments. However, existing solutions for blockchain scalability have limitations of either being increasingly cumbersome in security analysis or inherent deficiencies (e.g., surviving on duplicate transactions). Moreover, current state-of-the-art scalable blockchains suffer from low throughput when used for larger transaction blockchains. To improve scalability, we propose sibcha, a novel protocol that equipped with k (power of 2) parallel sibling chains that correspond to k transaction pools (indexed by the rightmost log₂k bits of transaction payers’ addresses). In the protocol, i-th transaction (along with a Merkle tree path) would be announced to the i-th chain based on the rightmost log₂k bits of the hashing determined in solving proof-of-work (PoW) puzzle (i is the exact value in decimal format represented by the log₂k bits). To achieve parallel transactions, we design a inter-chain mechanism without other correlations (such as block ordering, inter-chain transactions, block updates, eventual atomicity decoupling, two-phase PoW puzzle solving, etc.), which makes sibcha considerably simpler than current state-of-the-art solutions (e.g., OHIE at IEEE S&P 2020 and Monoxide at USENIX Security 2019). SibCha has much less (e.g., 1.86×∼ 3.16×) confirmation latency than OHIE. Prototype implementations also demonstrate that its throughput scales linearly with available bandwidth (1.5× that of Conflux).