Utilization vs. SLO-Based Control for Dynamic Sizing of Resource Partitions Zhikui Wang, Xiaoyun Zhu, Sharad Singhal HP Laboratories Palo Alto HPL-2005-126(R.1) January 23, 2006* server virtualization, resource partition, system identification, adaptive control In this paper we deal with a shared server environment where the server is divided into a number of resource partitions and used to host multiple applications at the same time. In a case study where the HP-UX Process Resource Manager is taken as the server partitioning technology, we investigate the technical challenges in performing automated sizing of a resource partition using a feedback control approach, where certain input variable such as the CPU entitlement for the partition is dynamically tuned to regulate output metrics such as the CPU utilization or SLO- based application performance metric. We demonstrate the importance of obtaining proper models to characterize both the static and dynamic input-output relations, identify the nonlinear and bimodal properties of the models across different operating regions, and discuss their implications for the design of the control loop. We then present various controller designs for either the relative utilization or the mean response time, evaluate the performance of the closed-loop systems while varying certain operating conditions, and discuss their advantages and issues. Finally, we present an adaptive controller that combines the CPU entitlement and utilization information and achieves more robust performance than prior solutions. * Internal Accession Date Only Published in and presented at the 16 th IFIP/IEEE Distributed Systems: Operations and Management (DSOM 2005), 24-26 October 2005, Barcelona, Spain Approved for External Publication © Copyright 2005 IEEE