A new self-adaptive extendible hash index for flash-based DBMS

Traditional magnet disk is not likely to improve read and write speed, due to its physical characteristics. Solid State Drive (SSD) has much higher random write and read speed than a magnet disk, and the absence of moving parts also makes SSD consume less energy and noiseless. However, traditional database can´t exert the high performance of SSD because of the different characteristics between magnetic disk and SSD. And flash-based index techniques are needed to overcome the high maintenance cost of traditional index. In this paper, we propose a new extendible hash index for flash-based DBMS, and a Split-or-Merge (SM) factor is added to make it self-adaptive. The analysis and experimental results show that our design minimizes the cost of index maintenance, and the SM factor makes it work efficiently for different write/update ratio.