As governments’ credit information platforms increasingly adopt permissioned blockchains, scalability requirements for permissioned blockchains have grown. Permissioned blockchains like Hyperledger Fabric and FISCO BCOS typically rely on crash fault tolerance (CFT) protocols such as Raft, due to their strict access controls. However, Raft faces scalability and replication challenges in permissioned environments, particularly with the resource-limited leader and a fixed replication model that struggles with fluctuating workloads. To address these, we propose PBRaft, a scalable, workload-adaptive variant of Raft. PBRaft improves scalability by alleviating the leader bottleneck and introduces a dynamic replication method to handle workload changes. Our implementation of PBRaft shows up to 25 \(\times \) higher throughput than Raft in extensive experiments.