Deterministic concurrency control method and system based on pre-transaction processing

Abstract

The invention provides a deterministic concurrency control method and a system based on pre-transaction processing, and the method comprises the steps: fully improving the expandability and performance of transaction processing on the basis of deterministic transaction processing through enabling the execution of a transaction to be before the scheduling of the transaction; ensuring the correctness of affair execution through verification and re-execution; satisfying the deterministic execution of the affairs through explicit waiting, and each affair is ensured to be re-executed once at most;using multi-version data storage, so that read-only transactions are prevented from blocking transaction processing; by recording the address of the data required by the transaction, accelerating there-execution after the transaction verification fails. Compared with the prior art, the method has the advantages that the concurrency of transactions can be fully utilized, and the transaction processing capacity of a deterministic database is greatly improved; data information collected during transaction execution can be reused to accelerate transaction scheduling; and the complete read-write set of the transaction is determined without an additional mechanism, so that additional performance overhead in the prior art is avoided.

Description

technical field [0001] The present invention relates to the field of distributed concurrency control, in particular to a deterministic concurrency control method and system based on pre-transaction processing. Background technique [0002] In order to cope with the explosive data growth of the Internet, distributed databases are widely used in different fields, and the requirements for database performance and distributed scalability are getting higher and higher. In recent years, the concept of deterministic databases has been proposed. This type of database system predetermines the global execution order for transactions, and ensures that transactions are executed in this order on each server. Compared with the traditional distributed database, each server of the deterministic database can independently reach a consensus on whether to commit the transaction, and does not need to use the distributed commit protocol to ensure the atomicity of the transaction commit, avoidin...

AI Technical Summary

Problems solved by technology

At the same time, deterministic locks need to apply for fine-grained locks for all read and write sets of transactions, and the index overhead of locks is large, which further leads to low performance of single-thread scheduling

[0007] 2. Limit the concurrency of transactions

Deterministic locks use locks to ensure that all transactions that may conflict cannot be executed in parallel. Therefore, this conservative scheduling method causes transactions that could have been parallelized to have to wait for one party, resulting in a decline in overall transaction processing performance.

[0008] 3. Transaction types are limited

Therefore, the previous work on the transaction scheduling of deterministic databases is not comprehensive, and there is still a certain gap

Images