The full-replication data storage mechanism, as commonly utilized in existing blockchains, is the barrier to the system's scalability, since it retains a copy of entire blockchain at each node so that the overall storage consumption per block is $O(n)$O(n) with $n$n participants. Yet another drawback is that this mechanism may limit the throughput in permissioned blockchain. Moreover, due to the existence of Byzantine nodes, existing partitioning methods, though widely adopted in distributed systems for decades, cannot suit for blockchain systems directly, so that it is critical to devise new storage mechanism for blockchain systems. This article proposes a novel storage engine, called BFT-Store, to enhance storage scalability by integrating erasure coding with Byzantine Fault Tolerance (BFT) consensus protocol. The first property of BFT-store is that the storage consumption per block can be reduced to $O(1)$O(1) for the first time, which enlarges overall storage capability when more nodes attend the blockchain. Second, we design an efficient online re-encoding protocol for storage scale-out and a hybrid replication scheme to enhance reading performance. Analysis in theory and extensive experimental results illustrate the scalability, availability and efficiency of BFT-Store via the implementation in an open-source permissioned blockchain Tendermint.

中文翻译：

---

许可区块链的可靠存储分区

现有区块链中普遍使用的全复制数据存储机制是系统可扩展性的障碍，因为它在每个节点保留了整个区块链的副本，因此每个块的总体存储消耗为 $O(n)$哦(n) 和 $n$n参与者。另一个缺点是这种机制可能会限制许可区块链的吞吐量。此外，由于拜占庭节点的存在，现有的分区方法虽然在分布式系统中广泛采用了几十年，但不能直接适用于区块链系统，因此为区块链系统设计新的存储机制至关重要。本文提出了一种称为 BFT-Store 的新型存储引擎，通过将纠删码与拜占庭容错 (BFT) 共识协议集成来增强存储可扩展性。BFT-store 的第一个特性是每个块的存储消耗可以减少到$O(1)$哦(1)这是第一次，当更多节点加入区块链时，这将扩大整体存储能力。其次，我们设计了一种用于存储横向扩展的高效在线重新编码协议和一种用于提高读取性能的混合复制方案。理论分析和广泛的实验结果通过在开源许可区块链 Tendermint 中的实施说明了 BFT-Store 的可扩展性、可用性和效率。