Method and system for optimizing B + tree index performance based on persistent memory

Abstract

The invention discloses a method for optimizing B + tree index performance based on a persistent memory. A traditional B + tree index is designed according to the characteristics of a DRAM and a disk,the characteristics of long writing delay, limited writing times, inconsistent power-down data and the like of a persistent memory are not considered, and the performance of a direct migration indexcan be greatly reduced. In order to improve the index concurrency performance, reduce the writing frequency of a B + tree in a persistent memory and prolong the service life of the persistent memory,the method adopts an additional writing strategy for leaf nodes to perform delayed refreshing on internal nodes, and reduces a large amount of shift writes brought to maintenance of entry orderlinessin the insertion process; in order to further improve the indexing efficiency, lock-free search is adopted in internal nodes, multiple threads can access the same node at the same time, and the indexconcurrency is improved.

Description

technical field [0001] The invention belongs to the technical field of database indexing, and more specifically relates to a method and system for optimizing the performance of a B+ tree index based on a persistent memory. Background technique [0002] Due to its high node fan-out, balanced tree height and dynamic resizing, B+ tree index is suitable for random and sequential processing application environments, has efficient data access performance, and is widely used in database systems. [0003] However, the traditional B+ tree index has some defects that cannot be ignored: first, it is designed according to the characteristics of DRAM and disk, and does not consider the characteristics of persistent memory such as long write delay, limited write times, and power-off data inconsistency. There are a large number of additional write operations. If the traditional B+ tree index is directly applied to the persistent memory, the index performance will be greatly reduced and the...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the above defects or improvement needs of the prior art, the present invention provides a method and system for optimizing the performance of B+ tree index based on persistent memory. When it is applied to the persistent memory, the index performance will be greatly reduced, and the technical problem of reducing the service life of the persistent memory, and in the multi-threaded mode, the existing B+ tree index mechanism has a negative impact on the locking mechanism of the shared data. Technical issues with large lock overhead

Images