Kim Kyung Hwan, a master's student in the Department of Computer Science, Graduate School of General Studies, conducted a study to improve the stability of spot instances by utilizing various spot instance-related data provided by cloud vendors.

Based on this, he presented a paper in the Oral Session of the Research Track of The Web Conference (WWW) 2024, the best academic conference in the field of computer science recognized by the National Research Foundation of Korea (acceptance rate: 9.4%). 

In his paper (Making Cloud Spot Instance Interruption Events Visible), he analyzed spot stability datasets provided by major cloud providers such as AWS, Azure, and GCP and showed that stable spot instances can be selected by predicting changes over time. The paper demonstrated that the proposed method can improve the stable uptime of spot instances by up to 168% compared to existing methods. This research on improving the stability of spot instances is crucial for the construction of cost-effective cloud environments and can be widely used in various research and industrial fields such as building various web services, AI learning and inference, and genomic information analysis.

Kim Kyunghwan is currently in the third semester of his master's program after graduating from the Department of Software at the College of Software Convergence, and is conducting research related to building performance and cost-effective cloud environments at the Distributed Data Processing System Lab (https://ddps.cloud) in the Department of Computer Science. The research on next-generation cloud systems that autonomously provide optimal AI service environments, including this research, is supported by SW Star Lab (Cloud), which is funded by the Ministry of Science and ICT and the National Institute of Information and Communication Planning and Evaluation.