Research On Multi-dimensional Metadata Management Based On Correlation Property

With the arrival of the cloud computing and cloud storage, the information storagesystem scale is expanding continuously. It is increasingly difficult on information storage,retrieval, analysis and processing. To efficient find and manage files in large-scale storagesystems, we need to use the file metadata to complete complex queries. Themulti-dimensional metadata information of the file system, such as file size, file type,access time and modification time is very important in the storage system. By analyzingand using the correlation properties of multi-dimensional metadata between different files,systems can be able to efficiently manage metadata and support complex queries.In large-scale storage systems, correlation properties refer to the aggregation effect offiles in its attribute space. It shows the correlation to a certain extent between the file data.Most metadata management systems, based on the analysis of the correlation properties,are mainly concentrated on the single dimension such as time or space, lack of analysis onmulti-dimension, making these systems not support complex metadata queries efficiently.Thus, using a simple, fast method to analyze, manage and recognize metadata attributes onmulti-dimension, and using the correlation properties between files, will directly improvethe efficiency of the access to the metadata.A metadata query system based on the multi-dimensional correlation properties offiles is designed and implemented. By using the locality sensitive hashing algorithm, thesystem clusters the similar metadata into same groups, creates efficient metadata indexand provides fast, accurate and complex metadata query service.The experimental results show that the system is able to support complex metadataqueries fast and accurately, provide a variety of query support for large-scale storagesystem on metadata with good scalability.