We are thrilled to present an enlightening interview with Dr. Minyi Guo , a visionary in the field of cloud computing. Renowned for his groundbreaking work on Self-Adaptive MapReduce (SAMR) and his contributions to the advancement of cloud-native technologies, Dr. Guo has made a profound impact within the industry. In this captivating discussion, we explore the motivations behind his innovative research, the influence of his work on industry giants like Microsoft Azure, Aliyun (Alibaba Cloud), Google, and Meta, and his insights into the future of cloud computing.
Join us in this interview as Dr. Minyi Guo shares his invaluable perspectives on the field, its transformative potential, and the key lessons he has learned throughout his illustrious career.
My research is driven by the heightening demand of providing high-quality online data-intensive computing services. For example, as China continues to build and expand its high-speed railway (HSR) system, online searches for train tickets grows rapidly. In addition, on the 11th day of every November, Alibaba’s global shopping festival, as called “double-11 shopping festival”, can attract hundreds of millions of consumers in a very short presale period. In the early 2010s, the services both faced saturated bandwidth, high latency, or even system failure.
When I first joined Shanghai Jiao Tong University, cloud computing was a new concept. I anticipated that the adoption of cloud computing would expand, and efficiently utilizing warehouse-scale computing resources can be a great undertaking. Therefore, we started to explore cloud computing to meet the quality-of-service (QoS) of many online services and to improve the sustainability of the underlying data center infrastructure.
Honor your colleagues achievements. Nominate Someone for a Major Award Today!
Before working on cloud computing, I worked on parallel and distributed computing for decades. When cloud computing appeared, I observed its potential impact in industry and believed that it would be the future of parallel and distributed computing. I also observed a large amount of new research problems in this new direction. For instance, I find that the classic straggler problem in distributed computing still hurts the overall performance in the traditional MapReduce, despite its easy programming and scheduling pattern. I therefore proposed SAMR to resolve the straggler problem, based on my experiences on distributed computing research.
I do expect to see my work influence large cloud vendors. When Aliyun (Alibaba Cloud) was first founded, I collaborated with it to help solve technical issues. For instance, I found that Aliyun is not capable of supporting the “double-11 shopping festival” due to the bursty load in a short interval. My research group focused on resolving such challenging problems for the full layers of the cloud computing system. Aliyun successfully resolved the problem by applying our techniques. While large cloud vendors usually face similar problems, my work is useful for them as well. By publishing my work, it would influence other large cloud vendors. Recently, I started to work on the fundamental container, serverless computing, and microservice scheduling topics in cloud native, and collaborate with Microsoft Research Asia (MSRA) on related topics. I believe that these techniques could also be used to the main cloud vendors like Azure, Google, and Meta.
What kept me motivated was the fact that energy efficient design has not gone out of fashion and has not become a consolidated field, with only incremental innovation. The challenge keeps rising, and we need to be more and more inventive, working across levels of abstraction.
To my understanding, there are two main potentials in cloud computing. The first potential is the transparent adaption to various heterogeneous architectures. While the current cloud native paradigm has enabled the fine-grained sharing and scheduling of resources, it does not support the heterogeneous devices very well. Current large cloud players are developing their own chips for better performance and total cost of ownership. However, the cloud system should be updated to adapt to these new chips. By providing a uniformed way to use X86 chips, Arm chips, RISC-V chips, even GPU and AI accelerators, we can greatly improve the efficiency of proving computing power to the tenants in clouds. Moreover, while artificial intelligence (AI) become more and more powerful, AI for cloud and cloud for AI also show great potentials. Specifically, emerging cloud applications are often empowered with AI capabilities, and have divergent requirements on the cloud resources. It is possible to utilize AI to better manage the cloud resources, and improve the cloud system to better support AI-based applications.
I believe it is important to interact with the industry and to touch emerging products. Most of my favorite designs have incorporated insights from my experience with production systems. I wish I could have received more feedback from my industry partners in my early career. Today, I often tell my students that high-quality papers are not the only source of inspiration; we have much to learn from the real world and then solve the real problems.
Dr. Minyi Guo ( 过 敏意 ) is the Zhiyuan Chair Professor, an IEEE Fellow, and an ACM Distinguished Member. Additionally, he is the Department of Computer Science and Engineering Director of the Embedded and Pervasive Computing Center at Shanghai Jiao Tong University. Dr. Minyi Guo received the BSc and ME degrees in computer science from Nanjing University, China; and the PhD degree in computer science from the University of Tsukuba, Japan. Before joined SJTU, Dr. Guo had been a professor of the school of computer science and engineering, University of Aizu, Japan. Dr. Guo received the national science fund for distinguished young scholars from NSFC in 2007. His present research interests include parallel/distributed computing, compiler optimizations, big data and cloud computing. He has more than 400 publications in major journals and international conferences in these areas. He received 7 best/highlight paper awards from international conferences including ALSPOS 2017 and ICCD 2018. He is now Editor-in-Chief of IEEE Transactions on Sustainable Computing, and on the editorial board of IEEE Transactions on Parallel and Distributed Systems, IEEE Transactions on Cloud Computing and Journal of Parallel and Distributed Computing. Dr. Guo is a fellow of IEEE, a fellow of CCF, and a distinguished member of ACM.
