Research On Transaction Concurrency And Prototype System Development Based On Hyperledger Fabric

Blockchain technology has gained widespread attention as an emerging technology in recent years.Hyperledger Fabric is an alliance blockchain platform.Due to its modular architecture,it is used to develop enterprise-level business applications.It is essentially a decentralized distributed database and a more secure and transparent transaction record plat-form than traditional database systems.However,compared to traditional databases,the Hy-perledger Fabric blockchain platform has a large gap in the speed of transaction processing,especially when processing transactions with concurrency conflicts,its processing efficiency is lower.To this end,this paper studies the concurrency performance of the platform and proposes an improvement plan,and finally implements a transaction prototype system.The main work of this paper includes the following three parts(1)This article proposes the LMLS method for the "write-write" concurrency conflicts existing on the Hyperledger Fabric blockchain platform.This method introduces a lock mechanism in the transaction process to enable some conflicting transactions to be marked at the beginning of the transaction process.Otherwise,indexes are added to conflicting transactions to optimize the storage of the ledger.(2)This article proposes a RWL method and a CM method to improve the "read-write"concurrency conflicts on the Hyperledger Fabric blockchain platform.The former uses read-write locks to ensure the correct execution order of conflicting transactions;the latter effec-tively improves data processing by introducing caches and ensures the consistency of the ledger and cached data by caching logs efficiency.(3)Based on the above two improvements,this paper uses the Hyperledger Fabric blockchain platform as the underlying architecture to design a prototype system that can handle highly concurrency transactions.This system can implement functions such as user transfers,deposits and withdrawals,and check balances.The system has a good concurrency capability.Experiments on a large number of data sets show that the methods proposed in this pa-per can greatly improve the platform's concurrency capabilities,and is suitable for different application scenarios with a good scalability.