Performance Characterization and Bottleneck Analysis of Hyperledger Fabric

Abstract

Hyperledger Fabric is a popular open-source project for deploying permissioned blockchains. Many performance characteristics of the latest Hyperledger Fabric (e.g., performance characteristics of each phase, the impacts of ordering services, bottleneck and scalability) are still not well understood due to the performance complexity of distributed systems. We conducted a thorough performance evaluation on the first long term support release of Hyperledger Fabric. We studied the performance characteristics of each phase, including execute, order, and the validate phase, according to Hyperledger Fabrics new executeorder-validate architecture. We also studied the ordering services, including Solo, Kafka, and Raft. Our experimental results showed some findings as follows. 1) The execution phase exhibited a good scalability under the OR endorsement policy but not with the AND endorsement policy. 2) We were not able to find a significant performance difference between the three ordering services. 3) The validate phase was likely to be the system bottleneck due to the low validation speed of chaincode. Overall, our work helps to understand and improve Hyperledger Fabric.

EngraveChain: A Blockchain-Based Tamper-Proof Distributed Log System

“…At the same time, it must be noted that RAFT is not robust against malicious attackers in a byzantine setting, in which case using PBFT would be necessary and there could be a performance penalty. A preliminary assessment [38] of the performance of the RAFT consensus found a throughput of several hundred TPS, which differed little from the performance using the solo orderer. For a fixed blockchain logic, the key performance factor was the endorsement policy, i.e., how many peers/organizations had to approve a transaction.…”

“…These benchmarks must be taken with some caution as they are based on a single orderer node, as in our implementation, however when combined with the work in Wang et al [38], we can be optimistic that changing to the RAFT consensus algorithm is unlikely to significantly reduce the performance.…”

“…Performance references for hyperledger fabric. (a) The transactions per second based on an endorsement policy requiring either of the two organizations to approve transactions (OR Policy)and bath of the organizations to approve the transactions (AND Policy)[38]. (b) The variation in performance based on the size of the asset created in the transaction[39].…”

QiOi: Performance Isolation for Hyperledger Fabric

“…In [19,48], the authors compared different consensus protocols such as Solo, Kafka, and Raft. Yusuf and Surjandari [19] demonstrated that Raft is superior to Kafka.…”

“…Yusuf and Surjandari [19] demonstrated that Raft is superior to Kafka. Wang and Chu [48] investigated the performance characterization of each phase, including execute, order, and validate phase. The authors found that the validate phase can become a bottleneck due to the low validation speed of chaincodes.…”