152 Supporting Efficient Query Processing on COB⁺ -Tree Available to Purchase

The effectiveness of query processing techniques on in-memory index data structure draws recently more and more attention, especially based on an compressed index. Index node compression improves data locality of index structure, decrease memory references. It makes node group sizes are different. However, the side effect of it is we cannot perform binary search on it. And decompressing can increase computational overhead. The contribution of this paper is three-fold. Firstly, we present a COB+-tree (Compression-Oriented CSB+-tree) data structure, which reduces the number of the second-level cache misses and the translation look-aside buffer misses. Secondly, we describe a fine measure of index compressing performance. Finally, by adopting partial decompressing technique, we can improve query processing performance on a COB+-tree. The experimental results show that the COB+-tree can improve the performance of query processing.