Node Compression Techniques Based on Cache-Sensitive B+-Tree

Abstract

Cache-conscious indices have been researched for a decade. These index structures can enhance data locality and reduce cache misses. Cache-Sensitive B+-tree (CSB+-tree) is a state-of-the-art, high performance index tree for main-memory database systems. However, in the worst case, the space utilization of CSB+-tree can fall to 25%. The main contribution of this paper is to improve space efficiency of CSB+-tree by using compression mechanisms. We have conducted experimental studies. Results show that the compressed CSB+-trees can yield significant space savings with additional benefits for search and delete operations by utilizing the superiority of modern hardware.