University of Michigan – Ann Arbor
Akshitha Sriraman is a Ph.D. candidate in the Computer Science Department at the University of Michigan. She is advised by Professor Thomas F. Wenisch on her computer architecture and systems dissertation research, specifically on the topic of enabling hyperscale web services. Her work bridges computer architecture and software systems and demonstrates the importance of that bridge in improving the performance and cost efficiency of modern hyperscale data center systems. Sriraman has influenced the design of server architectures both via hardware analysis of production data center systems and her subsequent software design that uses data center hardware more efficiently; she received recognition for this work by being nominated as a 2019 Facebook Fellowship Finalist. Additionally, Sriraman has developed a novel software system that improves data center performance by minimizing tail latency in large-scale systems; she was awarded the Rakham Merit Ph.D. Fellowship to help fund this work. She received her B.S. in Electrical Engineering from VTU, India in 2012 and her M.S. in Embedded Systems From the University of Pennsylvania in 2015.
Modern hyperscale data center web services such as web search, social networking, and software-as-a-service must often meet stringent performance objectives to satisfy end-user experience. At the same time, web service operators seek to maintain cost-efficiency through fungible hardware resources and inexpensive software techniques. Unfortunately, existing web systems introduce trade-offs between performance and cost-efficient data center system operation, often paying an expensive price to meet performance goals. My work addresses these challenges through practical and scalable systems that reconcile the conflict between performance and cost-efficiency in modern data centers. More specifically, I have analyzed the hardware requirements of large-scale production services to design a software system that automatically uses available inexpensive hardware more efficiently to improve web service throughput. I have also studied the performance implications of threading and concurrency design choices to develop a novel software system that automatically configures web service threading to improve performance. Moving forward, I aim to design holistic systems that improve performance and cost efficiency in the context of modern data center heterogeneity, sophisticated networking, and cloud security.