Transaction-redo-based multi-copy consistency maintaining method for heterogeneous clusters

Abstract

The invention relates to a transaction-redo-based multi-copy consistency maintaining method for heterogeneous clusters. The method is characterized by comprising the following stages: a query executing stage: a cluster transaction management module tracks an activity transaction list and records transaction operations, and clusters produce transaction redo logs for a master data node and send transaction redo log files to all copy data nodes and manage the states of the nodes; a copy maintaining stage: the copy data nodes receive the transaction redo log files from the master data node; and a failure recovery stage: database administrators find out archived logs corresponding to failed nodes and recover data of all the copy data nodes to a state consistent with that of the master data node. The method has the advantages that the problems of high cost and poor performance in the prior art are solved, the shortage that a binary log transmission mode cannot support heterogeneous databases is made up, the quick consistency maintenance among multiple copies is realized, the high availability is guaranteed in the case that no shared clusters exist, and meanwhile, the heterogeneity of cluster database nodes can be supported.

Description

technical field [0001] The invention belongs to the technical field of databases, in particular to a transaction redo-based multi-copy consistency maintenance method for heterogeneous clusters. Background technique [0002] Large-scale enterprise applications based on distributed database systems are becoming more and more widespread. In order to improve system reliability, the key data in the system will be stored in multiple copies in different places or locally, and the data in multiple copies must maintain consistency. Consistency includes strong consistency and weak consistency. Strong consistency requires data to be consistent at any time, that is, real-time consistency of data; weak consistency does not require real-time consistency of data, but Under certain conditions, data consistency is guaranteed, which is the so-called eventual consistency. The cost of maintaining strong consistency is relatively high. Therefore, it is generally used for key data with a small ...

AI Technical Summary

Problems solved by technology

This method can realize the multi-copy maintenance function to a certain extent in the homogeneous cluster, but it has obvious defects: due to the distributed architecture, the order of statements issued at the cluster layer and the actual execution order on the data nodes are inconsistent. Ensure consistency, so this method must serialize DML statements at the cluster middleware level, otherwise, it is likely to cause data inconsistency, or even serious problems such as the entire cluster deadlock

Obviously, this method is only applicable to the case of homogeneous database clusters, because the log formats of heterogeneous databases are bound to be different, and multiple copies cannot be maintained through log redo

Images