Data reading method and device

Abstract

A data reading method and device. The data reading method comprises: during a process of scanning an index clustering table, scanning leaf nodes and recording page numbers of the leaf nodes (S10); and merging I / O requests of the leaf nodes with consecutive page numbers into one I / O request (S11). Because data of multiple leaf nodes in physically consecutive storage locations is read by means of one I / O request, the data volume read by each I / O request is increased, thereby fully utilizing the overall disk throughput. In addition, to read data of the same storage volume, the number of I / O requests that need to be initiated can be reduced, thereby avoiding the limitation of the disk IOPS parameter, and effectively improving the data reading efficiency.

Description

technical field

[0001] The invention belongs to the technical field of computers, and in particular relates to a data reading method and device. Background technique

[0002] In relational databases, two-dimensional tables are usually used for data storage. Two-dimensional tables usually have two implementation methods, one is the implementation of the heap table, and the other is the implementation of the index cluster table. Among them, the heap table stores data in the free space in an unordered manner; the index cluster table is stored in the B-tree structure according to the order of the index keys.

[0003] The B-tree of the indexed cluster table uses a page as a unit as a node, and uses a page number to locate a page. Nodes include three types, root (root) nodes, branch (branch) nodes and leaf (leaf) nodes. The data of the database is stored in the leaf nodes. In the process of reading data, it is necessary to scan the index cluster table. The process of scanning ...

Images