Responsive and efficient provisioning for multimedia applications

Multimedia applications (including those in eHealth scenarios) can require on-demand and urgent resource provisioning in cloud environments. Provisioning in clouds, the virtual machines (VMs) assignment to physical machines (PMs), is critical to obtaining efficiency for the cloud provider and also to ensuring the overall satisfaction of cloud users. Provisioning algorithms are often divided into batch and online algorithms where the former gather VM allocation requests and then efficiently pack them as a group and the latter place VMs immediately upon receiving requests. The underlying tradeoff is one of efficiency in packing (leading to lower cost for the cloud provider) vs. greater responsiveness to requests (which is important to some cloud applications). In this paper, we propose and show the effectiveness of a new static hybrid algorithm for provisioning that groups and optimizes normal VM requests but immediately places more urgent requests.