Hierarchical Mixed Integer Programming for Pack-to-Swad Placement in Datacenters: Concept and Analysis

Computational clouds have evolved to go beyond cost-effective on-demand hosting of IT resources and elasticity. The added features now include the ability to offer entire IT systems as a service that can quickly adapt to changing business environments. The new trend introduces challenges in datacenter resource management, including scalability, system-orientation, and optimization supporting both datacenter efficiency and customer system agility and performance. The conventional VMcentric approach of resource management adopts a flat finegrained model that results in problem formulations of enormous sizes. The difficulty in scalability hinders cloud providers’ ability to improve resource efficiency and workload performance. In this paper, we introduce a pack-centric approach to datacenter resource management by abstracting a system as a pack of resources and considering the mapping of these packs onto physical datacenter resource groups, called swads. The assignment of packs/VMs to swads/PMs is formulated as an integer optimization problem that can capture constraints related to the available resources, datacenter efficiency and customers’ complex requirements. Scalability is achieved through a hierarchical decomposition method. The new datacenter resource management framework is illustrated with a concrete resource allocation problem, which is interesting by itself and challenging. The problem formulation, our solution, and numerical experiments show the capability of integer programming formulations, the scalability of the hierarchical decomposition method, and the benefits of the overall framework.