ScholarWorks@숙명여자대학교

초록

This study proposes a novel algorithm that transitions the allocation of hardware resources such as CPU, storage, and memory, as well as containers, in cloud environment from a traditional planar coordinate-based placement to a probabilistic mapping based on a spherical coordinate [1] system. To achieve efficient clustering of resources with diverse performance characteristics, the algorithm maps the dynamic attributes of resources to a probabilistic distribution on the sphere and performs resource allocation with optimal cost. This approach enables the formation of an ideal task clustering that considers both resource characteristics and dynamic system changes.

키워드