Research On B~+-tree Index Based On Dynamic Node Popularity

With the development of information technology and the arrival of the Web2.0,the type and scale of data resources expand unceasingly. Social development hasentered the era of big data, which brings great opportunities and challenges for indextechnology and the development of database. B+-tree index is an efficient way toindex orderly data, which plays an important role in creating index on file system anddatabase system. However, with the increasing of the data size, B+-tree node splittingspends a lot of time, and the internal node space utilization is low, thus affecting theperformance of application system seriously. Consequently designing efficient andreasonable database index structure to manage and organize these growing dataappears very necessary.This paper first introduces the common several kinds of index structure,including linear table index, order table index, hash index and tree index,emphatically expounds the B-tree index and B+-tree index. On the basis of thecomparison of the advantages and disadvantages of classical B+-tree mechanism andcombined with the current big data era background, due to the classic B+tree indexmechanism of query performance degradation and low space utilization, this paperproposes a modified B+-tree index structure-DNCPB+-tree, which is based on thedynamic poluparity of the node. This index structure improves the space utilization bymodifying the internal node structure of the classical B+-tree, and improve the queryperformance by introducing a parameter popularity on leaf node and dynamic cachehigh access frequency leaf node information in the internal node. Finally experimentsuse the SQLite open source database platform to compare the average I/O overheadand query response time so that we can verify the effectiveness of the DNCPB+-treeindex mechanism.The experimental results demonstrate that compared with the classical B+-treeindex mechanism, although DNCPB+-tree search will have additional overhead, it stillhas better query efficiency, and on average increased by18%, at the same timeimprove the space utilization.