Research On Efficient Temporal Data Query In Hyperledger Fabric System

In the traditional transaction scenario,all parties to the transaction need a centralized third-party to ensure the normal progress of the transaction.Bitcoin was born as a decentralized peer-to-peer digital currency system,which solved the trust problem to a certain extent.The underlying technology of Bitcoin is the blockchain,which is also considered as a distributed ledger.Due to its natural traceability and non-tampering characteristics,it is favored by various industries represented by the financial industry and supply chain industry.As a blockchain with traceability as the main application scenario,it is particularly important to query and analyze historical data in the time dimension.However,current blockchain solutions do not support temporal data processing,and sequential access based on block files in the blockchain hinders efficient query processing.In view of the low efficiency of temporal data query in the hyperledger fabric blockchain,this paper proposes two index building schemes.The first method proposes to use time interval segmentation on the historical data in the time dimension,determine the time interval in different ways,index the events at the same time interval,and establish the index according to the slant of the transaction data to reduce the number of blocks Access to files and databases improves the efficiency of temporal queries.The second method is for transactions with a large amount of data.In order to reduce the impact on the size of the block file itself,the file is used as the main storage,and the block transaction data arranged in chronological order is stored at a certain time interval to index the file.And use memory as a data buffer to speed up file reading and improve the efficiency of querying temporal data.In the experimental part,the thesis compares the query time on two datasets and analyzes the query performance.Experimental results show that the two methods proposed in this paper perform differently on different data sets,and are superior to other methods in overall time performance.