| Blockchain,the technology underpinning Bitcoin,is perceived to revolutionize various domains.Blockchain system has the advantages of decentralization and data tamper-proof,which has broad development prospects and diversified application sce-narios.Among many blockchain systems,Hyperledger Fabric(HLF)is one of the most commonly used blockchain systems,with the highest throughput of open source blockchain systems.However,there is still a certain gap between its performance and the demand of industrial application scenarios.In order to further expand and deepen the application scenarios of blockchain,it is of great practical value and significance to study and optimize the performance of Fabric blockchain system.The performance of HLF system includes single-chain transaction performance and multi-chain transaction performance.For single-chain transactions,existing HLF opti-mization methods optimize multiple processes,but there is still a lack of optimization for block submission process.This process can only be executed serially,and other pro-cesses are blocked during execution.Therefore,optimizing the efficiency of this process can improve the overall performance of the system.However,for multi-chain transac-tions,most of the existing research focuses on how to establish a reliable connection be-tween multi-chains,ignoring the performance of multi-chain transactions.The current multi-chain transaction model has the problems of simplicity and low efficiency.Multi-version Concurrency Control(MVCC)can manage transactions efficiently and improve performance.Therefore,by accelerating the block submission process of the HLF sys-tem and building a multi-chain transaction model based on MVCC,this thesis optimized the performance of the HLF system.The main research contents include:1.To propose cache-based performance optimization for single blockchain sys-tem:Aiming at the problem of low efficiency of block commentment in HLF,the re-source consumption and delay of each step of block commentment are analyzed in detail.On the basis of detailed analysis,a block cache method is proposed to solve the prob-lem of resource waste caused by redundant operations,and an index cache method is proposed to solve the problem of long index write delay2.To propose MVCC-based performance optimization for multi-chain blockchain system:Considering that Fabric doesn’t support multi-chain transactions,this thesis constructed a multi-version Merkle tree to support multi-version concurrency control of multi-chain transactions.In order to improve the efficiency of reading data in multi-version concurrency control,a fast snapshot matching scheme is proposed based on bloom filter.3.To build a high-performance blockchain platform prototype for supply chain fi-nance:In order to plan the development of blockchain applications,relevant standards have been developed.To meet the requirements of financial blockchain standards,a prototype of high-performance blockchain platform is constructed.This platform pro-vides reliable enterprise credit information and financial financing services for supply chain finance,and achieves good results. |
PDF Full Text Request