Lock function implementation method and device and computer medium

Abstract

The invention relates to the technical field of distributed databases, and particularly provides a lock function implementation method which comprises the following steps: S1, establishing a lock controller in a memory to perform single-point management on distributed transaction concurrence, ensuring the consistency through a ramp synchronization mechanism, and implementing lock table recovery through transaction logs redo and undo after node downtime; s2, a lock and an exclusive lock are shared to give the read transaction to obtain the right of the lock so as to block other concurrent read transactions and write transactions; and S3, mapping the write intention of the MVCC into a memory lock table, namely a copy lock, so as to ensure that a lock controller adapts to MVCC timestamp control. Compared with the prior art, the method has the advantages that conflict judgment of a storage layer is improved to a memory, the concurrency performance of the distributed database is improved, the influence of the read request on the timestamp of the write request is avoided, and meanwhile application scenes are greatly enriched.

Description

technical field [0001] The invention relates to the technical field of distributed databases, and specifically provides a lock function realization method, device and computer medium. Background technique [0002] The distributed database based on the percolator transaction model uses multi-version timestamps (MVCCTimestamp Ordering) for concurrency control, and uses write intents to ensure that data consistency is not destroyed when transaction conflicts occur, so as to ensure that the database receives concurrent transaction requests. Ensure transaction consistency and isolation (ACID principles). [0003] For multi-version timestamp control, each write operation will create a version of the data for the data item. When the transaction issues a read operation, the concurrency control manager will select a version from the data item to read, which can ensure Reads are not blocked by writes, and writes are not blocked by reads. When a transaction conflicts, it is necessary...

AI Technical Summary

Problems solved by technology

[0008] Since the multi-version data of MVCC is persistent data stored in the disk, not in the memory, this ensures that the transaction does not cause conflicts during the concurrent process, and requires a large number of reads and writes to the disk, resulting in performance degradation of the database. Decrease, because access to disk data is much more expensive than access to in-memory data

Images