Design and Implementation method of a high concurrent index B + linked list data structure

Abstract

The invention discloses a design and implementation method of a high concurrent index B + linked list data structure. The method comprises the following steps: simultaneously using an array-based datastructure and an element-based data structure; for the non-leaf nodes of the data structure, the preset B + tree data structure is adopted and placed in the DRAM, so that the non-leaf nodes located on the DRAM can guarantee the locality of access; for the leaf nodes of the data structure, a unidirectional linked list data structure is adopted, which is placed in the NVM, so that the leaf nodes located on the NVM can avoid sorting and balancing overhead. The method uses array-based data organization and element-based data organization, Linked list data structure constructs the leaf node of index data structure, B + tree data structure constructs the inner node of index data structure, skip table data structure removes sorting and balancing operation, which can realize lock-free concurrencymechanism and effective space management, and ensure efficient concurrency access performance and rapid system recovery.

Description

technical field [0001] The invention relates to the technical field of non-volatile main memory storage, in particular to a design and implementation method of a highly concurrent index B+ linked list data structure. Background technique [0002] Non-volatile main memory (Non-Volatile Memory, NVM) is a new type of memory storage medium, which has byte addressability, non-volatile information after power failure, high storage density, no need for dynamic refresh, and static power consumption. low merit. At the same time, there are also some shortcomings, such as asymmetric read and write performance, limited write times and high write power consumption. Its emergence has brought great new opportunities and challenges to the storage field, and has triggered a research boom in the industry and academia on heterogeneous hybrid memory architecture and related system software. Non-volatile memory has many new implications for computer system architecture, system software, softwa...

AI Technical Summary

Problems solved by technology

The main reason for these problems lies in the traditional B+ tree, which still adopts an array-based structure for the organization of nodes. The sorting and balancing problems caused by this structure make the writing cost relatively high. This cost requires additional maintenance of faults. In the case of consistency, it will become larger and thus cause a large write amplification overhead

Secondly, the coarse-grained array structure will be locked at the granularity of the whole, and the high-overhead sorting and balancing operations will further increase the duration of holding a lock. This overhead will be more reflected in the larger NVM media write delay. obvious

[0004] In addition, some data structures optimized for NVM will lead to multi-version garbage of version elements and low utilization of tree nodes, which will lead to serious space utilization problems

At the same time, the memory allocator for NVM will also bring some performance degradation

Images