Seminar: Identifying and Understanding Performance Problems in Software Systems

 

  

Fang Zhou

Meta Platforms Inc.

Tuesday January 24, 2023

10:30 am

Location: Patrick F. Taylor Hall Room 1212

Abstract

Performance analysis is fundamental and significant to software development in both academia and industry. Especially in the era of post Moore’s law, performance analysis becomes increasingly important and challenging due to stagnant per-core CPU performance and slow increase of core count. In addition, computer systems have become more complex with the emergence of numerous new hardware and innovative software techniques. As a result, improving the performance of computer systems is more challenging.

In this talk, Fang will present his work on performance analysis to identify bottleneck waiting events. Then he will introduce the trace-based emulation approach to understand performance problems in a complex system. In the end, he will share his future work on performance analysis and operating systems.

 

Bio

Fang Zhou is a research scientist at Meta Platforms Inc. He received his Ph.D. in Computer Science from The Ohio State University in 2021. His primary research interests include performance analysis, operating systems, and distributed systems. He has published several papers in system conferences including OSDI, EuroSys, and VLDB. He is an AEC member of Journal of Systems Research.