CAREER: Low Latency I/O and Ubiquitous Storage
Princeton University, Princeton NJ
Investigators
Abstract
The role played by magnetic storage systems is becoming increasingly important as computing systems accelerate their transformation from "number-crunching" devices to "information appliances". However, the technology has advanced so far that the inadequacies of the state-of-art are becoming apparent. The first of these is the lack of understanding of the science behind the architecture of a device that delivers optimal performance/cost ratio. The second inadequacy is the lack of coordination of the various levels of hardware and software in a storage system. The third is that the file system designers have not yet recognized the emergence of the ubiquitous consumer devices as one of the network storage research issues. This research explores two critical questions: how to provide very low latency I/O and how to manage ubiquitous network storage. Low latency is achieved by a combination of innovations in architecture of the devices themselves and exploiting the interactions between these devices, file systems, and applications. Ubiquitous storage management is achieved by leveraging the emergence of ubiquitous connectivity and a peer-to-peer architecture. By addressing the issues of performance and functionality of magnetic storage systems, the proposed research contributes to the science underlying a crucial backbone of the information infrastructure.
View original record on NSF Award Search →