Abstract：To solve the bottleneck caused by metadata management subsystem in cloud storage, a metadata caching algorithm of cloud storage was proposed. Based on the characteristics analysis of accessed metadata, the concept of metadata caching life cycle was determined. The calculating rules of metadata caching life cycle was designed, and the metadata caching algorithms of metadata substitution and writeback were presented based on the life cycle definition. The metadata management efficiency of cloud storage was greatly improved by the caching algorithms. The metadata caching algorithms adaptability to user accessing characteristics was analyzed, and the metadata consistency guaranteed by the algorithms was discussed. A metadata caching prototype was realized based on the proposed algorithms. Analysis and experiments were completed on the benchmark dataset by general test tools. The results show that the proposed algorithms can effectively enhance I/O rate by 15% and operation processing rate by 16% in cloud storage, respectively.
2012, Vol. 33 
