Michael McThrow: Metadata Management

Student's Name: 
Michael McThrow
None
Advisor's Name: 
Scott Brandt
Home University: 
Cal Poly San Luis Obispo
AttachmentSize
Image icon Mike_McThrow.jpg64.04 KB
PDF icon mcthrow.pdf122.54 KB
Image icon mcthrow_picture1.jpg31.9 KB
PDF icon mcthrow_paper.pdf107.26 KB
Year: 
2006

As file systems have to deal with a growing amount of files and file formats, the need for file systems to manage extended metadata has grown. Existing file systems are not equipped to handle managing metadata, leaving applications to put their file-based metadata within their files. The problem with that approach is that the format of the metadata is often proprietary and hard to access with other applications. This makes it difficult to search for files based on their metadata.

What is the best way to store this extended metadata? We hypothesize that relational databases are the best way to store extended metadata in a file system. Relational databases are a proven tool for storing data, and relational databases have over three decades worth of improvement and optimization. Relational databases are also queryable. By making a file system queryable using relational databases and queries, searching for files based on their metadata becomes an easier and more intuitive process.