Toward QoS provision for virtualized resources in grids

Abstract

Virtualization solutions have recently gained considerable attention for supporting the consolidation of application servers into one physical machine. In a vanilla Xen implementation the scheduler shares equally all of the available physical CPU resources among the contending VMs. However, when the application that runs in the virtual machine space changes dynamically its resource requirements, a different solution is needed. Furthermore, if the resource usage is associated with service-level agreements, a predefined equal share of the processor power is not sufficient for the VMs. In this chapter, we present an approach to manage the QoS of virtualized resources in Grids. Our solution adjusts the resources needed by each VM according to an agreed QoS. We achieve this by a local resource manager (LRM), which we implemented as a prototype and deployed on Xen-virtualized machines. By means of experiments we show that the implemented management component can meet the service-level objectives (SLOs) by dynamically adjusting virtualized resources according to demand.