Dictionary symbol encoding: A column-store case study

摘要

The performance of disk I/O for large data transfer is driven by compression ratio. In this paper, we studied and explored the compression of large column-store databases and proposed the organization of the compressed data on the concept of dictionary encoding logical compression scheme in matrix form. Our experiment shows that with proposed compression algorithm, TPC-H query performance may improve up to 21% approximately and space consumption may reduce by 2%.