Dynamic Hilbert curve-based B+-Tree to manage frequently updated data in big data applications

Abstract

In big data application sets, the values of the data used change continually in practice. Therefore, applications involving frequently updated data require index structures that can efficiently handle frequent update of data values. Several methods to index the values of frequently updated data have been proposed, and most of them are based on R-tree-like index structures. Research has been conducted to try to improve the update performance of R-trees, and focuses on query performance. Even though these efforts have resulted in improved update performance, the overhead involved and the immaturity of the concurrency control algorithms of R-trees render the proposed methods a less-than-ideal choice for frequently updated data. In this paper, we propose an update-efficient indexing method. The proposed index is based on the B+-tree and the Hilbert curve. We present an advanced Hilbert curve that automatically adjusts the order of the Hilbert curve in sub-regions, according to the data distribution and the number of data items. We show through experiments that our strategy achieves a faster response time and higher throughput than competing strategies.