The Hashing Approach to the Internet File System Problem
In here we consider the problem of developing distributed file system over the Internet. Such a file system should efficiently simulate a shared hard disk created from many local disks over the Internet. The key feature of the proposed Internet File System (IFS) is that the mapping between files and physical bocks on the various machines is based on hashing and not on a global search tree as is in the case with regular file-systems. Consequently any client around the glob can directly compute the "physical" address (Internet Address) of a block in a file and update it. The lack of a global data structure to locate physical blocks yields an extremely simple file-system through which many clients can perform remote read/write/open/close/delete operations over the Internet.
We have implemented the system over a cluster of PCs running Windows2000 and used a video player to test its performance. In this respect our work also contributes to the well-known problem of VOD (video on demand). Our results suggest that hashing based distributed file systems can be extremely efficient for driving Internet file systems and file systems for large clusters.