Proportional Allocation of Resources on Shared Ring Buffer for Virtualization

Abstract

In virtualization, without proper scheduling and resource management, a load surge in one VM may unacceptably degrade the performance of another. Key requirement of IO performance virtualization is performance isolation. The current state of performance isolation for virtualization is more rudimentary. This paper presents a resources allocation framework, based on abstracted Xen IO model, that provides both throughput and fairness guarantees for network and storage device via 2-level resource control. The high-level tier makes guest domains perceive the states of the resource through resource control on sharing ring buffer. The high-level tier uses public/private token-bucket to support work-conserving and fairly distributes any spare bandwidth to the different VMs. The low-level tier is intended to meet the fairness guarantee, and computes token quantity value of each bucket by feedback-driven scheduler. The authors analytically and experimentally demonstrated that their framework is work-conserving, and achieves fairness, and high adaptability.