A data migration method and device for distributed database cluster

Abstract

The invention discloses a data migration method and device of a distributed database cluster. The cluster comprises m database nodes and n fragments. Each database node is responsible for storing data of one or more fragments. The method includes the steps that according to a set distribution mode, data entering the cluster is distributed to the corresponding fragments; when data migration is needed in the cluster, the database nodes needing data migration are determined; the fragments, needing to be migrated, in the fragments managed by the determined database nodes are selected out; data of the selected fragments is migrated to the specified database nodes from original database nodes. The data migration scheme supports various scenes like single-node expansion data migration and cluster overall expansion data migration, and universality is high.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a data migration method and device for a distributed database cluster. Background technique [0002] With the rapid development of Internet technology, users generate more and more data. The reliable storage and use of these massive data has become the key to computer technology. In terms of technology, it is necessary to ensure the storage security of massive data and the availability of these data, so that users of massive data can use data efficiently and transparently. Distributed systems provide solutions for the storage and use of massive data. As a massive data storage in a distributed system, an important problem to be solved is to determine the data distribution strategy in the cluster. A good distribution strategy should be able to evenly distribute data to all nodes, and should also be able to adapt to changes in cluster nodes . It can be seen that data mig...

AI Technical Summary

Problems solved by technology

[0011] 1. When the overall data volume of the cluster increases and needs to be expanded, adding a node only shares the data volume of one node in the original node, and cannot balance the whole; this will cause unbalanced data distribution and introduce access bottlenecks to the entire system. It cannot solve the problem of overall expansion; doubling all nodes brings waste of resources;

[0012] 2. When a single node performs data migration for capacity expansion, it needs to rehash all the data on the original node, and the performance is obviously low;

[0013] 3. It cannot support complete data migration across computer rooms due to computer room upgrades, etc.

Images