Experience management for task placements in a cloud

Abstract

The execution of workflows in a cloud is more and more popular, and new business concept based on this combination emerge. However, the task to control a cloud in such a way, that the rented cloud resources match the requirements for the currently executed workflows is difficult. Simple solutions struggle with over-, and under-provisioning problems or lack the needed flexibility for the new business concepts. A smart concept for cloud management should use knowledge about the characteristic of the executed task to improve the resource utilization of the cloud. In this paper we present our approach for a CBR based concept for cloud management that reuses experience on proper cloud configurations. We introduce our similarity function for task placements in a cloud and illustrate the approach with some sample workflows form the music mastering domain.