Resource Management in Utility and Cloud Computing

We are entering an era of “Everything-as-a-Service” where resources are shared at an unprecedented scale. The so called utility computing model, built upon cloud computing infrastructures, becomes ubiquitous in the enterprise IT landscape. In this chapter, we first introduce recent advances in the study of the economics of cloud computing, known as Cloudnomics. Next, we describe the motivation behind cost-effective resource management design in Cloudnomics. We also summarize relevant research to the study of resource management in utility and cloud computing. At the end of this chapter, we describe the fundamental research challenges, present our design evolutions with regard to these challenges, and sketch our proposed solutions.

Application services using cloud computing infrastructure are proliferating over the Internet. In this chapter, we study the problem of how to minimize resource rental cost associated with hosting such cloud-based application services, while meeting the projected service demand. This problem arises when applications incur significant storage and network transfer cost for data. Therefore, an Application Service Provider (ASP) needs to carefully evaluate various resource rental options before finalizing the application deployment. We choose Amazon®; EC2 marketplace as a case of study, and analyze the optimal strategy that exploits the tradeoff of data caching versus computing on demand for resource rental planning in cloud. Given fixed resource pricing, we first develop a deterministic model, using a mixed integer linear program, to facilitate resource rental decision making. Next, we investigate planning solutions to a resource market featuring time-varying pricing. We conduct time-series analysis over the spot price trace and examine its predictability using Auto-Regressive Integrated Moving-Average (ARIMA). We also develop a stochastic planning model based on multistage recourse. By comparing these two approaches, we discover that spot price forecasting does not provide our planning model with a crystal ball due to the weak correlation of past and future price, and the stochastic planning model better hedges against resource pricing uncertainty than resource rental planning using forecast prices.

In this chapter, we investigate the resource trading problem in a utility and cloud computing setting where multiple tenants communicate in a Peer-to-Peer (P2P) fashion. Enabling resource trading in cloud unleashes the untapped cloud resources, thus presents a flexible solution for managing resource allocation. However, finding an efficient and fair resource allocation is challenging mainly due to the heterogeneity of resource valuations. Our work first develops a utility-oriented model to support resource negotiation and trading. Based on this model, we adopt a multiagent-based technique that allows a group of autonomous tenants to reach an efficient and fair resource allocation. Further, we add budget limitation to each tenant and propose a directed hypergraph model to facilitate resource trading amongst heterogeneous tenants. We develop a directed hypergraph model to facilitate trading decision making, and design a class of heuristic-based distributed resource trading protocols in favor of different performance metrics.

This chapter presents CloudBay, an online resource trading and leasing platform for multi-party resource sharing. It is a proof-of-concept design bridging the gap between resource providers and resource customers. With the help of CloudBay, the untapped computing power privately owned by multiple organizations is unleashed. The design and implementation of the CloudBay project presents the most challenge to our exploration of cost-effective resource management strategy design. Following a market-oriented design principle, CloudBay provides an abstraction of a shared virtual resource space across multiple administration domains, and features enhanced functionalities for scalable and automatic resource management and efficient service provisioning. CloudBay distinguishes itself from existing research and contributes in mainly two aspects. First, it leverages scalable network virtualization and self-configurable virtual appliances to facilitate resource federation and parallel application deployment. Second, CloudBay adopts an eBay-style transaction model that supports differentiated services with different levels of job priorities. For cost-sensitive users, CloudBay implements an efficient matchmaking algorithm based on the auction theory and enables opportunistic resource access through preemptive service scheduling. The proposed CloudBay platform stands between HPC service sellers and buyers, and offers a comprehensive solution for resource advertising and stitching, transaction management, and application-to-infrastructure mapping. In this chapter, we present the design details of CloudBay, and briefly discuss lessons learnt and challenges encountered in the implementation process.

We have explored the design space for cost-effective and flexible resource management strategies in utility and cloud computing. In this final chapter, we summarize our findings and discuss related future work directions based on the solutions depicted in this book.