Optimizing Long-term IaaS Service Composition
We propose a new economic model based optimization approach to compose an optimal set of infrastructure service requests over a long-term period. The service requests have the features of variable arrival time and dynamic resource and QoS requirements. A
- PDF / 383,469 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 81 Downloads / 157 Views
Abstract. We propose a new economic model based optimization approach to compose an optimal set of infrastructure service requests over a long-term period. The service requests have the features of variable arrival time and dynamic resource and QoS requirements. A new economic model is proposed that incorporates dynamic pricing and operation cost modeling of the service requests. A genetic optimization approach is incorporated in the economic model that generates dynamic global solutions considering the runtime behavior of service requests. Experimental results prove the feasibility of the proposed approach.
1
Introduction
Cloud computing is increasingly becoming the technology of choice as the nextgeneration platform for conducting business. Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS), and Software-as-a-Service (SaaS) solutions have already been offered by big companies in the cloud market [1]. An IaaS provider receives service requests in the form of computing resource (i.e., functional) and Quality of Service (QoS) (i.e., non-functional) requirements. The provider associates resource specifications (e.g., CPU, Memory, and Network Bandwidth) and QoS attributes (e.g., availability, throughput, and response time) with provided Virtual Machines (VMs) [2]. For the latter, the provider usually specifies them in a Service Level Agreements (SLA), where SLA violations may incur penalties for the provider [1]. The key challenge for the provider is the optimal composition of the service requests that closely meet the provider’s economic expectation by considering certain constraints, such as resource limits and SLA violations. The provider-consumer relationship between IaaS and SaaS providers is longlasting and economically driven [3]. The contract period of a long-term service are usually counted in months or years. Due to the nature of the cloud consumers, e.g., multi-tenancy and changing business and cost requirements, a fixed set of requirements are often inapplicable in a long-term period. The key characteristic of the long-term service requests is that its functional and non-functional requirements change from time to time [3]. The long-term economic expectation of an IaaS provider also can change over time. Here we consider profit, SLA violations, and resource utilization as the key components in the long-term economic expectation. For example, a provider may find that SLA violations in the summer period influence the reputation more than the other periods. In this c Springer-Verlag Berlin Heidelberg 2015 A. Barros et al. (Eds.): ICSOC 2015, LNCS 9435, pp. 333–342, 2015. DOI: 10.1007/978-3-662-48616-0 22
334
S. Mistry et al.
research, we assume that the long-term economic expectation of an IaaS provider has already been defined. The objective is to select an optimal set of long-term service requests that satisfy the provider’s long-term economic expectations. In our previous research, we propose a new model for predicting dynamic consumer request behavior in long-term IaaS service compositions
Data Loading...
