A New Multi-Agent Hybrid Marketplace for Cloud Resource Allocation
- PDF / 3,987,977 Bytes
- 50 Pages / 439.37 x 666.142 pts Page_size
- 99 Downloads / 192 Views
A New Multi‑Agent Hybrid Marketplace for Cloud Resource Allocation Sepideh Adabi1 · Vida Esmaeili1 Received: 7 June 2019 / Revised: 27 January 2020 / Accepted: 13 February 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Stagnation of the negotiation-based cloud market refers to a condition of having negligible or no successful (good) deals. Cloud market stagnation should be reduced as it has a detrimental effect on the market outcomes. One of the effective steps towards overcoming market stagnation is by focusing on temporary replacing the pricing mechanism with the more effective ones. To do this, we propose a new agent-based hybrid marketplace in which the elements of the negotiation market are combined with the elements of auction as another well-known market-driven pricing mechanism. That is, negotiation market is considered as the main basis of the proposed hybrid market where each resource provider can make decisions about holding ad-hoc auction(s) when experiencing high market pressure to exit the market stagnation and create dynamism in the market. Thus, in this paper, issues and challenges related to when and how an auction should be held are answered. Extensive amounts of simulation were carried out to evaluate the performance of dealer agents who deal in the proposed hybrid market in comparison with MBDNAs known as (Market-and-Behavior-Driven Negotiation Agents) in terms of average utility, success rate, and the time needed to reach an agreement. The results show that our designed dealer agents outperform MBDNAs. Keywords Cloud computing · Market-based resource allocation · Dynamic pricing · Auction model · Negotiation model · Multi-agent system · Fuzzy Inference System
* Sepideh Adabi Sepideh_adabi@iau‑tnb.ac.ir Vida Esmaeili [email protected] 1
Department of Computer Engineering, North Tehran Branch, Islamic Azad University, Tehran, Iran
13
Vol.:(0123456789)
Journal of Network and Systems Management
1 Introduction Cloud computing is emerging as a new computing paradigm. Customers hire their required resources based on the pay-as-you-go model so they are charged only for whatever they use. Cloud computing services can be divided into three main categories: Software as a Service (SaaS), Platform as a Service (PaaS), and Infrastructure as a Service (IaaS) [1]. IaaS is the foundation layer of cloud computing which presents computational infrastructure as a service to end users. IaaS provides a way for users in order to monitor, manage and hire resources by developing VM instances on physical resources. As stated in [2]: Amazon EC2 [3], Eucalyptus [4], Nimbus [5], OpenStack [6], and Open Nebula [7] are examples of cloud infrastructure implementation. IaaS cloud providers rent out their resources by adopting different pricing strategies. In the cloud business model, several resource customers can compete for hiring the same resource and also several cloud providers can compete to rent out their resources to the resource customers. Two well-known strate
Data Loading...
