Blockchains offer trust and immutability in non-trusted environments, but most are not fast enough for latency-sensitive applications. Hyperledger Fabric (HF) is a common enterprise-level platform that is being offered as Blockchain-as-a-Service (BaaS) by cloud providers. In HF, every new transaction requires a preliminary endorsement by multiple mutually untrusted parties called organizations, which contributes to the delay in storing the transaction in the blockchain. The endorsement policy is specific to each application and defines the required approvals by the endorser peers (EPs) of the involved organizations. In this paper, given an input endorsement policy, we studied the optimal choice to distribute the endorsement requests to the proper EPs. We proposed the OPEN algorithm, devised to minimize the latency due to both network delays and the processing times at the EPs. By extensive simulations, we showed that OPEN can reduce the endorsement latency up to 70% compared to the state-of-the-art solution and approximated well the introduced optimal policies while offering a negligible implementation overhead compared to them.

翻译：区块链在非信任环境中提供了信任与不可篡改性，但大多数区块链对延迟敏感型应用而言速度不足。Hyperledger Fabric (HF) 是常见的由云提供商以区块链即服务 (BaaS) 形式提供的企业级平台。在HF中，每笔新交易需先经多个互不信任的参与方（称为组织）的初步背书，这导致了交易存储至区块链的延迟。背书策略专属于具体应用，并规定了相关组织背书节点 (EP) 所需的批准。本文针对给定输入背书策略，研究了将背书请求分发至恰当EP的最优选择。我们提出了OPEN算法，旨在最小化由网络延迟及EP处理时间共同导致的延迟。通过大量仿真实验证明，与现有最优方案相比，OPEN可将背书延迟降低高达70%，且能良好逼近所引入的最优策略，同时相较这些策略仅带来可忽略的实现开销。