Market-Oriented Cloud Computing: Vision, Hype, and Reality for Delivering IT Services as Computing Utilities

This keynote paper: presents a 21st century vision of computing; identifies various computing paradigms promising to deliver the vision of computing utilities; defines Cloud computing and provides the architecture for creating market-oriented Clouds by leveraging technologies such as VMs; provides thoughts on market-based resource management strategies that encompass both customer-driven service management and computational risk management to sustain SLA-oriented resource allocation; presents some representative Cloud platforms especially those developed in industries along with our current work towards realising market-oriented resource allocation of Clouds by leveraging the 3rd generation Aneka enterprise Grid technology; reveals our early thoughts on interconnecting Clouds for dynamically creating an atmospheric computing environment along with pointers to future community research; and concludes with the need for convergence of competing IT paradigms for delivering our 21st century vision.

💡 Research Summary

The paper opens with a bold vision of the 21st‑century computing utility, arguing that the ultimate goal of IT is to provide computing resources on demand, just as electricity or water are delivered today. After reviewing earlier distributed paradigms—grid, cluster, peer‑to‑peer—the authors contend that these approaches lack the economic and service‑level guarantees required for a true utility. They therefore define cloud computing as a virtual‑machine‑based, highly automated platform that offers services at three abstraction levels (IaaS, PaaS, SaaS) and, crucially, introduce the concept of a “market‑oriented” cloud.

The market‑oriented architecture is layered. At the bottom lie physical hosts and hypervisors that host virtual machines (VMs). Above them sits a VM‑management layer responsible for provisioning, migration, and isolation. The top layer is the market mechanism, composed of a service broker, a dynamic pricing engine, an SLA management module, and a risk‑assessment component. The broker translates user QoS requirements into resource requests; the pricing engine computes real‑time prices based on supply‑demand fluctuations; the risk component quantifies the probability of SLA violations and feeds this information into a risk‑aware allocation algorithm. This design enables providers to maximize revenue while honoring contractual guarantees, and it gives customers transparent, usage‑based billing.

To illustrate the state of the art, the authors survey major commercial clouds—Amazon EC2, Google App Engine, Microsoft Azure—highlighting how each implements virtualization, auto‑scaling, and pricing. They note that while all three rely on VM or container abstractions, their business models differ: EC2 uses per‑hour instance pricing, App Engine adopts a fine‑grained “pay‑as‑you‑go” for compute cycles, and Azure blends IaaS with enterprise‑grade management tools.

The paper then presents the authors’ own research platform, the third‑generation Aneka enterprise grid. Aneka extends traditional grid scheduling with market‑driven price negotiation, allowing users to specify cost‑performance targets when submitting jobs. The system automatically selects a mix of local and cloud resources that satisfies the user’s constraints while optimizing provider profit. This hybrid approach demonstrates how grid and cloud technologies can be merged under a unified market‑oriented framework.

A forward‑looking section introduces “atmospheric computing,” a vision of inter‑cloud federation where independent clouds interconnect through standardized APIs, mutual authentication, and distributed SLA negotiation. In such an ecosystem, a user’s workload could seamlessly spill over from one provider to another during peak demand, achieving global elasticity and reducing idle capacity across the whole cloud fabric.

Finally, the authors outline open research challenges: rigorous economic validation of dynamic pricing models, formal methods for end‑to‑end SLA enforcement across multiple clouds, quantitative risk‑management models, and the development of policy and regulatory frameworks that support market‑driven resource allocation. They argue that only by converging the technical advances of virtualization, grid scheduling, and service‑level engineering with robust market mechanisms can the computing utility vision become a reality. The paper concludes with a call for joint academic‑industry efforts to build the standards, tools, and business practices needed for this convergence.