Zhang leads NSF-funded project to address computational storage

Posted: October 20, 2022

Computer Science and Engineering (CSE) Professor Xiaodong Zhang leads a multi-institution team that recently was awarded a $1.2 million grant from the National Science Foundation (NSF) to move computing power close to storage systems of increasingly high capacity.

The idea of empowering memory and data storage devices with an embedded and dedicated computing facility has been around for more

Xiaodong Zhang
Zhang in his Dreese Labs office

than 20 years with many published papers. However, Zhang and his team observed that storage industries could hardly gain benefits from the available academic research results for three reasons.

First, the available designs in research projects assumed that external bandwidth between compute nodes would always be lower than that of internal bandwidth inside a compute node, which has been proven to be incorrect. Second, the cost and power constraints were largely not considered in the academic research projects. Finally, the interface between the designed computational storage systems and existing computing systems was customized, which hindered it becoming a part of general-purpose computing ecosystems.

The team’s funded project, “A New Direction of Research and Development to Fulfill the Promise of Computational Storage,” will address these critical issues of computational storage and aim to develop the new system to be a dependable part of computing infrastructure in both software and hardware.

The other two PIs are Louisiana State University Associate Professor Feng Chen and Rensselaer Polytechnic Institute Professor Tong Zhang. The three PIs have a long-term collaboration track record on storage research with impact. For example, their advanced error correction code (ECC) implementation has been widely adopted in commercial solid-state devices (SSDs) in both customer-grades and enterprise-grades by major companies, such as Intel and Samsung. Their storage buffer management and hybrid storage designs have been adopted in the products of Apple and Oracle.     

“I am glad that the panel ranked our proposal to be highly competitive," said Zhang. "Moving computing power to storage systems is also new research to me. I am confident that our research will influence industries on the development of next generation storage systems.”

To systematically integrate computational storage into existing computing ecosystems, the research team will design a service-oriented abstraction for applications, optimize system-level resource utilization, and leverage proximity and mitigate memory resource contention in device hardware, among other tactics.

Zhang and his colleagues also aim to make a broader impact by training students at different levels with research activities, enriching curriculum and classroom teaching with new research results, and contributing to outreach activities. Zhang has consistently opened his lab to undergraduate students to do research, many of whom went to top CSE graduate programs at Ohio State and other universities including University of Illinois, University of Southern California, and New York University.

double-arrowFaculty Profile