Towards Optimizing Hyperledger Fabric Performance through Strategic Waiting

Hyperledger Fabric (HLF) is an open-source distributed ledger for businesses to establish private blockchains with granular control over data access and privacy. It typically has multiple endorsing, committing, and ordering peers that collectively contribute to the network’s functionality and security. For optimal performance, balancing the endorsement load among peers is essential. In practice, though, a subset of leaders often become solely responsible for endorsements, making the HLF performance sub-optimal. In this early work, we show that pausing commits at the current leading peers—to enable other peers to catch up and become leaders themselves—can lead to non-trivial throughput improvements. Even with a basic strategy, we see 1.27× more commits in synthetic experiments. This paves the way for sophisticated optimal commit strategies in future.