Early-concept Grant for Exploratory Research (EAGER) funds novel architectures for memory caching systems

The Department of Computer Science extends its heartiest congratulations to Anshul Gandhi and Erez Zadok who were recently awarded $257k in National Science Foundation (NSF) funding to investigate novel architectures for memory caching systems which will result in significant cost and energy savings for system administrators. 

Gandhi and Zadok’s research will be funded through NSF’s Early-Concept Grants for Exploratory Research (EAGER), Gandhi and Zadok. The EAGER program supports exploratory work in its early stages that has the potential to transform research ideas or approaches.

Realizing that the available RAM on a single computer is not enough for caching the amount of data used by distributed services today, many companies such as Facebook and YouTube utilize a distributed memory caching system called "Memcached". Memcached creates a cluster of nodes, forming a key-value database in faster memory. Memcached's simple architecture led to its popularity, but is also its Achilles' heel. Any changes to the number of Memcached nodes result in discarding most cached data, leading to lengthy periods where the distributed caches have to be slowly re-warmed up from the backend database. Gandhi and Zadok’s research seeks to alleviate this problem.

This EAGER project investigates techniques to scale Memcached clusters smoothly, without any transient performance degradations. The project also introduces an intermediate Flash-based storage tier between the memory nodes and the backend database, to help reduce high latencies to the backend database and help the smooth scaling of the Memcached cluster.

Both Gandhi and Zaodok are members of the Smart Energy Technologies group at Stony Brook. Their EAGER project is comprised of two one-year long phases. Throughout each phase, the team plans on developing tools to allow for full-system evaluation, and will even publicly release those tools, along with their workloads, data sets, and software. Release of the tools and workloads will fill the gap of available benchmarking tools for memory caching tiers.

Ultimately, the team hopes to provide efficient memory caching techniques to the entire computer science community. Their work will cut energy costs for data centers, benefitting the economy and their solutions apply to many large internet companies including, Amazon. The project, which began in May 2016 represents the eighth faculty member to receive an EAGER award in the Department of Computer Science, which is part of the College of Engineering and Applied Science.  

About the PIs

Anshul Gandhi serves as Principal Investigator (PI) on this EAGER project. His high-level areas of research including applying theory to systems and leveraging mathematical tools to analyze system behavior. He earned his PhD from Carnegie Mellon University and served as a Post-Doctoral Researcher in the Cloud Optimization and Analytics group at the IBM TJ Watson Research Center. Gandhi joined SBU in 2014 and earlier this year he was the recipient of a Google Faculty Research Award.

Erez Zadok, a Co-PI on this effort, focuses his research interests on file systems and storage, operating systems, energy efficiency, security, and performance and benchmarking. Zaodok completed his PhD in Computer Science at Columbia University and was recently named a full professor in the department. In addition to being part of the Smart Energy Technology group, Zadok directs the File Systems and Storage Lab (FSL). Zadok is the recipient of the SUNY Chancellor's Award for Excellence in Teaching, an NSF Career Award, and two IBM Faculty awards