The Case for Reexamining Integrated File System Design

Research in integrated file systems—file systems that support different types of data and applications with diverse performance requirements—has stagnated since the commercial failure of video-on-demand systems in the 1990s. However, the past two years have seen an acceleration in the development and deployment of several new technologies aimed at web-based multimedia applications. These technologies have opened up a fertile research ground for new file system optimizations and rekindled interest in integrated file systems. In this paper, we describe some of these technology trends, examine their implications on file system design, and present a sampling of the new research problems that arise due to these technological changes. Based on these observations, we argue for a renewed effort in designing next-generation integrated file systems. I. TECHNOLOGY TRENDS An integrated file system is a general-purpose file system that supports applications with diverse performance requirements (e.g., interactive, throughput-intensive, soft real-time) accessing data with different characteristics (text, images, streaming media). While integrated file systems have evolved from continuous media servers, they are inherently different and more complex than a videoon-demand server. Specifically, a continuous media server is a special-purpose file system optimized for a single data type, namely streaming media, while an integrated file system supports a diverse set of applications and data types found in general-purpose computing environments. An integrated file system is also different from a conventional file system—the former is aware of the semantics of the stored data and employs mechanisms that exploit these semantics to cater to different types of applications, while the latter is typically oblivious of semantics of files and provides a simple best-effort service to all applications. Whereas several integrated file systems were designed in the mid 1990s [4], [5], [6], [11], research in integrated file systems has stagnated in the recent past. However, the development and deployment of several new technologies for emerging web-based multimedia applications has neThis research was supported in part by an NSF CAREER award CCR-9984030, NSF grants ANI-9977635, CDA-9502639, Intel, Sprint, and the University of Massachusetts. TABLE I PROJECTED GROWTH IN NUMBER OF USERS WITH HIGH-SPEED INTERNET ACCESS Year Cable Modem DSL Wireless broadband 2000 1.6 million 0.78 million 52,00