Public Spot Instance Dataset Archive Service
Kyunghwan Kim, Subin Park, Jaeil Hwang, Hyeonyoung Lee, Seokhyeon Kang, Kyungyong Lee · 2023
Spot instances offered by major cloud vendors allow users to use cloud instances cost-effectively but with the risk of sudden instance interruption. To enable efficient use of spot instances by users, cloud vendors provide various datasets that reflect the current status of spot instance services, such as savings ratio, interrupt ratio, and instant availability. However, this information is scattered, and they require distinct access mechanisms and pose query constraints. Hence, ordinary users find it difficult to use the dataset to optimize spot instance usage. To resolve this issue, we propose a multi-cloud spot instance dataset service that is publicly available. This will help cloud users and system researchers to use spot instances from multiple cloud vendors to build a cost-efficient and reliable environment expediting cloud system research.