Flash-aware Clustered Index for Mobile Databases

Abstract

Flash memory become the very popular storage technology. Recently, it may be treated as a main storage memory in mobile devices, PDA and personal computers. However, the architecture based on flash memory has some limitations. They suffer from poor write performance, as the flash blocks must be erased before write. In particular, small random writes are very inefficient in comparison to read operations. Such asymmetry has implications as far as data management is concerned. Many database storage techniques must be changed to consider the new hardware characteristics. This paper proposes a new clustered index which considers the flash memory limitations. Due to utilizing fractional cascading and tree-like structure, the efficiency of update and search is obtained. The analytical and experimental results show that the proposed approach outperforms the traditional clustered index based on B+tree in terms of the flash memory limitations. The proposal may be useful in the mobile databases which operate on the flash memory.