Growth and Efficiency in the Cloud
For the past year ISyE Assistant Professor Martín Zubeldía and ISyE Ph.D. student Yishun Luo have been working together to find ways to improve the operational efficiency of data centers, sometimes also known as the Cloud. Originally spurred by Luo, the topic aligns with Zubeldía’s background in Electrical Engineering and Computer Science, as well as his passion for theoretical and algorithmic work.
“Luo was interested in how data centers operate and how to make them more efficient. In particular, he was interested in reducing how much computers need to communicate with each other without sacrificing performance. With this goal in mind, Luo developed a new load balancing algorithm that has computers communicate much less with each other and still achieve the same performance levels.”
Zubeldía likens their work on improving efficiency of data centers to the progression of the telephone. The first telephone network had people manually connecting cables to link different telephones, which was good enough for very small networks. This would not work today because of the sheer number of connections that we make every day, so telephone connections had to be made more efficient in order to meet the demands of the users. In the same way, the exponential growth in the demand for data center processing capabilities necessitates a similar increase in operational efficiency.
“The amount of data and service requests has grown exponentially, and older algorithms aren’t good enough because they don’t scale up appropriately. We have to keep up with the level of demand and develop more efficient algorithms in order to meet this growth in demand.”
These efficiency improvements not only appease the user, whether you are streaming, downloading or uploading images and videos, or are a gamer, but it also allows the companies to save money and energy, and invest in their users. Initially an ISyE master’s student, Luo’s enthusiasm for learning and his self-driven mentality was what Zubeldía finds most enjoyable about working with Luo. Zubeldía is very impressed with Luo coming up with this research direction, and with the new algorithm. This shows that Luo has the qualities that will make him an excellent Ph.D. student. Zubeldía is eager to continue his work with Luo and to see him grow as a researcher.
Luo explains that the idea for this project emerged almost serendipitously.
“As I was experimenting with simulations on various queuing models—exploring different approaches just to see what might work—a surprising result surfaced. With Professor Martín’s insightful guidance, this discovery evolved into the heart of our research. I am deeply appreciative of my time as a master’s student in our department, which allowed me to take Ph.D. level courses, broadening my understanding of advanced optimization and stochastic models with the guidance of our exceptional professors. Working alongside Professor Martín has been particularly rewarding; he has not only directed my research, but has also demonstrated the qualities of a committed, passionate researcher. His dedication to inquiry and mentorship has profoundly influenced my approach to research, and I look forward to learning even more from him as I continue my academic journey. Looking ahead, I’m excited to continue making meaningful contributions—pursuing research that drives our field forward with creativity and impact.”
For Zubeldía, his research and teaching are all about the theoretical and algorithmic tools that can be used broadly in many application domains. He knows that these tools are important for student growth, and can be applied to future projects in the students’ careers.
“To learn the tools and be able to apply them anywhere is important. You don’t know where you are going to end up after your Ph.D. - it could be in a finance company, and then at a company like Amazon. All of these companies have different issues, so if you have a good education in the fundamentals, then you can go to any company and make a positive impact. The application domain comes from the company, but we should be able to provide the solution.”