Multi-tenant utility computing with compute containers

From the growing demand for utility computing resources emerged both the hardware and software for cloud computing technology. Still varied enterprises to fulfill their capacity need while adhering a strict fiscal cost structure. However, the exponential increase in demand and complexity of the network topology, threatens to simultaneously violate both initial cost and capacity aspects. In order to best fulfill customer needs, cloud service providers (CSP) should operate together to create a scalable, novel, and fair ecosystem. While every CSP wishes to provide its value proposition, there is also a need for a cohesive utility service that enables generic computing resources. Such eco-system should be bottom line attractive financially and should also fulfill the elasticity and density properties of cloud computing. This paper explores a container-based paradigm for achieving dense and elastic virtualization technologies. It describes on-going experiments with several discreet Linux Containers (LXC) and illustrative applications. The experiments shows 22% improvements in CPU cores allocation when using compute-containers. It demonstrate the proven range of both the density and elasticity properties via a prototype with both virtualization paradigms that runs an illustrative distributed application. The prototype manifests a fine-grained resources allocation with the container paradigm and indicates its suitability, advantages and disadvantages for utility computing services.