A major challenge for any distributed file system like GlusterFS, is conducting data maintenance without seriously affecting I/O response time. Solutions like a namespace crawl are inherently inefficient and may not provide correct information for any single point in time. Using a small, well integrated database provides a major improvement over a namespace crawl, with minimal impact to system performance. The talk will focus on the results of a study to integrate a database with GlusterFS. Also the talk will include a demo.
Hall: D406 [4th Floor]
Track: Distributed Storage