Asaf Cidon, assistant professor of electrical engineering and of computer science, has won a National Science Foundation (NSF) CAREER award to build a new software framework that will speed up cloud computing and improve storage operations. The five-year, $508 thousand dollar grant, the NSF’s most prestigious award given to junior faculty to support early-career development, will support his proposal, “In-Kernel Execution of Storage Functions.”
Cidon, who joined Columbia Engineering in 2019, conducts research on software systems that power cloud computing. His group performs experimental research--from designing new types of storage and memory systems that support fast, large-scale big data analytics to designing machine-learning-based scheduling algorithms that reduce energy consumption in large data centers. With the emergence of microsecond-scale storage and persistent memory devices, he has concentrated his research on fundamentally rethinking large parts of the cloud's software infrastructure to efficiently exploit these devices.
Cidon’s NSF project is focused on building a new software framework, XRP, that would enable cloud applications using fast storage devices to offload commonly used storage functions, such as index traversals and aggregations, to the operating system. By doing so, XRP allows applications to eliminate the significant overhead of traversing all the operating system layers and going back and forth between the application and the operating system, each time the storage device is accessed. Cidon expects XRP to more than halve the amount of computation and energy needed to conduct common storage operations on fast storage devices. XRP would be supported by existing operating systems and file systems, thus requiring minimal changes from the application developer in order to be adopted, and reducing the barrier for real-world adoption.