INVITED SPEAKER SEMINAR - Profile-guided Optimization for Cloud Services: Accelerating Serverless Cold Starts and Reducing Unnecessary Service-to-Service Communication

Date: Tuesday, August 19, 2025, from 1:00 p.m. 

Location: EV 3.309

Abstract

Cloud service performance is critically influenced by efficient resource utilization and communication overhead. Cloud providers commonly optimize resource usage at the infrastructure level; however, inefficiencies introduced by application-level code often remain overlooked. Unlike developers in the high-performance computing (HPC) community, cloud service developers lack extensive experience in performance-oriented code optimization, necessitating specialized tool support explicitly tailored for optimizing cloud-native applications. This talk presents profile-guided optimization techniques specifically designed for cloud environments, addressing two key performance challenges: unnecessary inter-service communication in microservice applications and library loading overhead during serverless cold starts. First, we introduce MicroProf, a profiling tool that precisely attributes unnecessary data transfers to specific code-level interactions, thereby significantly enhancing the efficiency of microservice applications. Second, we present SLIMSTART, a runtime profiling solution that addresses serverless cold-start latency by dynamically identifying and mitigating redundant library initializations. Integrated seamlessly into existing CI/CD workflows, these scalable profiling tools deliver substantial improvements in end-to-end performance and resource utilization. While these tools provide deep insights, developers often find it challenging to translate profile data into actionable code optimizations. My current research now explores the use of large language models (LLMs) to automate and enhance code‑optimization tasks, laying the groundwork for future benchmarking refinements, tooling extensions, and open‑source innovations in performance engineering.

Biography

Dr. Probir Roy is an Assistant Professor in the Department of Computer and Information Science at the University of Michigan-Dearborn. His research focuses on building practical tools and techniques to expose various performance bottlenecks and optimization opportunities in software. Since the complexity of computational systems is increasing, designing performant software/hardware is becoming more challenging. To address these challenges, he develops practical "program analysis tools." He got his Ph.D. while working with Xu Liu at the College of William and Mary, developing the Lightweight Memory Profiling Techniques to Identify Performance Inefficiencies.