Proper

Abstract

The conventional hard disk has been the dominant database storage system for over 25 years. Currently, flash memories are one of the best media to support database storage areas. However, PRAM is better than flash memory in the respects of read and write performance. The growing column-oriented storage has become an attractive repository for big data warehouses because of its superior compression effects. Recently, hybrid storage which incorporate the advantages of PRAM memory into the flash memory are considered to be the next storage systems to support big database servers. However, we need to improve traditional index management schemes based on B-Tree due to the slow write characteristics of flash memory. In order to achieve this goal, we propose an index management scheme called SB+-Tree. Our SB+-Tree-based index management enhances update performance by caching node objects in shadow area of PRAM to reduce slow flash I/Os in index access processes.