Almost every filesystem makes a hierarchical abstraction of the data contained in the device. The many advantages of this presentation form are the reason for its popularity. However, it also has some restrictions and drawbacks. E.g. when a user wants to keep the same file in different locations, he has the possibility to do that with shortcuts (Windows) or Links (Unix). This might be an adequate solution as long as you don't need to move the actual location of the files. Copying is not the answer to our problem either since this could lead to incoherencies.

This is where our Virtual File Service comes into play. It makes an additional abstraction of the filesystem granting a relational view of the files. Links are no longer managed by the machine's filesystem but by the Virtual File Service. Therefore, the software is a distributed system and is split into an information server (InfoServer) and one or more fileservers (InfoClient). The clients can make use of the offered services. Thereby the information server is responsible for the abstraction of the filesystem. The software running on the fileserver (InfoClient) has to monitor the filesystem for relevant filechanges and notify these to the information server. The virtual view of the filesystem is no longer dependent on the location of the files. The communication between InfoServer and InfoClient is encrypted and requires the authentication of the InfoClient. The implementation of the userclient is not part of this project. Only an API is provided which provides an interface to the Virtual File Service. The authentication of the userclient is implemented with the kerberos authentication method.

Finally we hope that future users will benefit from our work on the Virtual File Service and that the hopefully soon implemented userclient might be a comfortable and new way to access the filesystem.