Efficient Materialized View Maintenance And Trusted Query For Blockchain

As a distributed ledger technology maintained by multiple parties,blockchain has recently been applied in many industries,such as financial.The characteristics of tamperproof,decentralization and traceability provide fundamental infrastructure support for building decentralized applications(Dapp).There exist research works focusing on the management of blockchain data due to the inconvenience and low efficiency of querying on blockchain.Some recent works that try to integrate blockchain and database have enhanced the ability to manage blockchain data.However,none of these efforts can improve the query performance on blockchain data.The reason why it's challenging to query block data is that transactions that belong to the same tables are stored across blocks.When applying materialized views,one of the common approaches to improve query performance in traditional database,to blockchain,the following three challenges need to be considered: 1.how to reduce the maintaining cost of materialized views is a key problem in this thesis.2.The proposed materialized view maintenance method needs to support bulk operations since different types of transactions in one block can trigger the maintenance of multiple views.3.Materialized views stored locally may be tampered with,hence certain measures have to be taken to ensure the correctness of query results.Considering all three challenges,this thesis proposes an efficient materialized view maintenance method that selects appropriate times for maintenance and adopts Merkle tree to guarantees the correctness of query results.The main contributions are listed below:1.To our knowledge,this thesis employs materialized views to blockchain for the first time and proposes an efficient maintenance algorithm.The algorithm chooses appropriate times for maintenance by considering the consensus process of blockchain.Our method supports bulk maintenance for a batch of transactions in one block as well.2.To deal with the security problem when storing views off-chain,we propose a query authentication approach based on Merkle tree to guarantee the correctness of query results.In advance,the Merkle-tree is constructed for the materialized view.When the query needs to scan the materialized view,a new Merkle-tree is constructed according to the data rows of the view.Finally,the roots of the two trees are compared to determine whether the materialized view is tampered.3.We implement our materialized views along with the view-based query and authentication process in a real blockchain system with relational semantics.The materialized view mechanism is well integrated into the blockchain system and improves the query performance of block data significantly.In summary,the integration of materialized views and blockchain enhances query efficiency and while introducing minimum impact on the system.