Snapshot-based data index in cloud storage systems

摘要

With the arrival of the big data era, the data in storage system becomes lager and lager. The size of data index also increasing in a high speed. It is a crucial problem that how to index data quickly in a distributed data storage system. Meanwhile, because of the increasing size of data index, the traditional way, which rebuilding the index or fully backup the index before the data is changed, cannot match the need of today's big data index. In this paper, a double-layer B+ tree based index structure that overcomes the throughput limitation triggered by single index server is proposed. Then a high efficient snapshot-based B+ tree structure is designed to adapt dynamic cloud environment. Finally, experimental results show our snapshot based solution is feasible and efficient.