By splitting a large index of complex objects into subsets and placing each of them on a separate machine, retrieval throughput would be improved invoking parallelism. We employ the multiindexing technique which exhibits good adaptability to index splitting. In this paper, we propose a parallel index retrieval scheme in which objects in output packets are integrated in order to increase the packet density. New retrieval termination detection algorithms are incorporated to adapt the packet integration.
Citation:
Takeshi Horie, Takuya Ukon, Tatsuo Tsuji, Ken Higuchi, "An Efficient Parallel Retrieval for Complex Object Index," icdew, pp.1268, 21st International Conference on Data Engineering Workshops (ICDEW'05), 2005