Implementing File Systems and Object Databases in a Microstorage Architecture

Abstract

A microstorage architecture consists of a microstorage kernel and several storage servers. Each storage server implements a storage model that defines a client's view of all the data in the system, how it is stored, retrieved and manipulated. The storage servers are built on top of the microstorage kernel and rely on it to perform the actual data storage and retrieval. The microstorage kernel implements a mechanism and the storage servers each implement specific policies defined by their storage models. Several different storage servers, each implementing a different storage model, may run concurrently over the microstorage kernel and all data in the system is concurrently visible to all the different storage servers. Different application programs, or different parts of the same application program, can use different storage models to manipulate the same data. Microstorage architectures provide a flexible interface and a smooth transition from traditional file systems to more powerful object oriented storage models. Existing applications continue to work correctly unchanged because they use a storage server that implements a traditional file model while new applications may gradually adopt more powerful storage models.