Elsevier

Decision Support Systems

Volume 52, Issue 3, February 2012, Pages 562-576
Decision Support Systems

Modeling and analysis of the effects of QoS and reliability on pricing, profitability, and risk management in multiperiod grid-computing networks

https://doi.org/10.1016/j.dss.2011.10.012Get rights and content

Abstract

In this paper we develop a network equilibrium model for optimal pricing and resource allocation in Computational Grid Network. We consider a general network economy model with Grid Resource Providers, Grid Resource Brokers and Grid Users. The proposed framework allows for the modeling and theoretical analysis of Computational Grid Markets that considers a non-cooperative behavior of decision-makers in the same tier of the grid computing network (such as, for example, Grid Resource Providers) as well as cooperative behavior between tiers (between Resource Providers and Grid Brokers). We introduce risk management into the decision making process by analyzing the decision-marker's reliability and quality of service (QoS) requirement. We analyze resource allocation patterns as well as equilibrium price based on demand, supply, and cost structure of the grid computing market network. We specifically answer the following questions with several numerical examples: How do system reliability levels affect the QoS levels of the service providers and brokers under competition? How do system reliability levels affect the profits of resource providers and brokers in a competitive market? How do system reliability levels influence the pricing of the services in a competitive environment? How do users' service request types, QoS requirements, and timing concerns affect users' behaviors, costs and risks in equilibrium? How does the market mechanism allocate resources to satisfy the demands of users? We find that for users who request same services certain timing flexibility can not only reduce the costs but also lower the risks. The results indicated that the value of QoS can be efficiently priced based on the heterogeneous service demands.

Highlights

► We develop model for pricing and resource allocation in grid computing network. ► We introduce risk management into the decision making process. ► We analyze the effects of reliability and quality of service (QoS) on the grid. ► Users' services time flexibility can reduce the grid costs and risks. ► Results indicate that QoS can be priced based on heterogeneous service demands.

Introduction

Grid computing is a form of distributed system wherein computing resources are shared across networks. These resources include CPU hours, storage, and applications and they are shared based upon their availability, capability, and cost, as well as the user's quality of service requirements. Consequently, grid technology can help organizations accelerate application performance, improve productivity and collaboration, and reduce total cost of IT infrastructure ownership. Grid computing has evolved from a niche technology associated with scientific and technical computing into a business-innovating technology that is driving increased commercial adoption. The growth of the grid computing market coupled with the increase in the number of market participants has generated many new challenges. In particular, since the grid-computing market is essentially a decentralized decision-making network, system reliability and quality of service (QoS) are among the major concerns of the market participants and users. In this paper, we study how the issues of quality of service, reliability, risk and standards can be resolved in this emerging market, and how these issues will affect the decision process and support systems as well as a company's bottom line. In particular, we investigate the following questions:

  • How do system reliability levels affect the QoS levels of resource providers and brokers in a competitive market?

  • How do system reliability levels affect the profits of resource providers and brokers in a competitive market?

  • How do system reliability levels influence the pricing of services in a competitive environment?

  • How do users' service request types, QoS requirements, and timing concerns affect users' behaviors, costs and risks in a competitive market?

  • How does the market mechanism allocate resources to satisfy the demands of users with different QoS requirements?

Recently grid computing market design and resource allocation have been a theme of many studies. A significant contribution to establish a market vocabulary and structure that can be associated with grid resource markets is from the work of Buyya [6], Wolski et al. [52], and Bapna et al. [3]. Wolski et al. [52] considers a multi-commodity (CPU and storage) resource market and suggests that a tātonnement-based commodities exchange market structure is a better choice for controlling grid resources than auction strategies previously defined in Regev and Nisan [41] and Waldspurger et al. [48], particularly with respect to price stability and resource utilization efficiency. Buyya [6], on the other hand, focuses on the implementation issues of scheduling and resource brokering and suggests a variety of market mechanisms, including models based on commodity markets, posted-prices, auctions and bargaining. Lin and Lin [34] examine the optimal service priority selection problem for a grid computing service user, who submits a multi-subtask job for the priced services in a grid computing network. Bapna et al. [3], however, were the first to model grid resource allocation based on economic principles. They develop a decentralized market model using a combinatorial auction approach and consider only two classes of decision makers, buyers and sellers. Zhang et al. [57] use real option valuation technique and simulation to analyze economic decision criteria for a grid computing provider wishing to provide such a service to businesses. These studies, however, did not consider QoS, reliability, and risk management in a multiple criteria, multiple resource, multiple decision-makers and multiperiod grid computing network optimization framework.

We propose what we believe is a novel approach to the modeling and analysis of grid computing markets. We model the multicriteria decision-making behavior of various decision-makers, which includes: the maximization of net profit and the minimization of risk. The proposed framework which is based on network equilibrium theory, allows for explicit modeling of decentralized decision-making behavior of the market participants. This approach differs from the recent models [3], [6], [48], [51], [52] in that, first and foremost, we consider several different types of decision makers and model their behavior and interactions explicitly. Second, we introduce risk management into the decision making process by analyzing the decision-marker's reliability and QoS requirement. Finally, the proposed framework provides the tools for computing efficient resource allocation patterns as well as equilibrium price patterns based on demand, supply, and cost structure of the grid computing network.

Many authors have previously addressed the issues of QoS, reliability, and risk in grids. In terms of QoS, Dogan and Ozguner [13] consider the problem of scheduling a set of independent tasks with multiple QoS requirements. Golconda and Ozguner [28] compare five QoS-based scheduling heuristics. Ernemann [15] addresses the idea of applying economic models to the scheduling task. Chunlin and Layuan [10] consider scheduling finite resources to satisfy the QoS needs of various grid users with multiple dimensional QoS requirements. In this paper, however, we take a network market approach to QoS and assume that the user's transaction cost is a function of QoS requirement. Moreover, we assume that QoS is a function of the individual decision-maker's system reliability.

In terms of reliability, Hwang and Kesselman [22] present a failure handling system based on work flows. Jin et al. [27] propose a framework for the adaptive deployment of failure detectors and, based on it, a policy-based failure handling mechanism to choose the appropriate failure recovery method. Kola et al. [32] provide a classification of faults in large distributed systems (with the main focus on grids). Huedo et al. [21] evaluate the reliability of computational grids from the end user's point of view. The book of Xie et al. [55] present many models for measuring reliability of grid computing system. In this paper, in addition to modeling the reliability of individual decision nodes and links, we incorporate reliability into the decision making processes.

Most of the research concerned with grid computing risk have addressed some form of risk in computing jobs. Irwin et al. [24] and Popovici and Wilkes [40] minimize the risk of paying penalties to compensate users so as not to reduce the profit of service providers. Kleban and Clearwater [30], [31] determine the risk of completing jobs later than expected based on either the makespan (response time) or the expansion factor (slowdown). Yeo and Buyya [56], in contrast, provide a way to evaluate reliability and risk with respect to the required objectives, such as the deadline to complete the job, the budget and the penalty for any deadline violation. Nevertheless, these research consider the case of a single decision-maker and did not consider the grid network market competition and risk.

The grid computing network market equilibrium approach, developed in this paper, permits one to represent the interactions between decision-makers in the market in terms of network connections, reliability, QoS, risk, flows, and prices. We consider the relationships among QoS, reliability and risk. In addition, we consider non-cooperative behavior of decision-makers in the same tier of the grid computing network (such as, for example, grid resource providers) as well as cooperative behavior between tiers.

The paper is organized as follows. In Section 2, we present the model setup. In Section 3, we model the various decision-makers and their behavior and analyze the equilibrium among the heterogenous decision makers. In particular, we develop a variational inequality formulation governing the entire grid computing network. In Section 4, we conduct computational studies to investigate the five questions raised in the beginning of the paper. In Section 5, we discuss the managerial insights. In the concluding Section 6, we summarize our results and suggest directions for future research.

Section snippets

Grid computing network market model setup

In this Section, we develop an equilibrium model for a grid computing network market in which decision-makers operate in a decentralized manner. We develop an economic model of the computational grid market that provides an optimal pricing and resource allocation mechanism. Grid computing is exemplified in the Globus project [19] and GrADS project [4]. We note that grid computing is closely related to the fasting emerging business model of “cloud computing”. In particular, Aymerich et al. [2]

The behavior of decision-makers and equilibrium analysis

We now turn to describing the behavior of the various economic decision-makers. We first focus on the resource providers. We then turn to the brokers, and, subsequently, to the users. An equilibrium solution is denoted by “*”.

Computational studies

In this section we investigate the interactions between QoS/reliability and the decision-making of the players in the computational grid network. In particular, we will utilize a series of numerical examples to study the following questions:

  • How do system reliability levels affect the QoS levels of resource providers and brokers under competition?

  • How do system reliability levels affect the profits of resource providers and brokers in a competitive market?

  • How do system reliability levels

Managerial insights

Our results reveal important insights regarding reliability, QoS, pricing, and profitability for policy makers, managers as well as users.

For policy makers, our analysis shows that in a healthy competitive market, decentralized decision-making can efficiently allocate resources with consideration of system reliability. In particular, if sufficient information regarding firm's reliability is accessible to market participants and users, the market mechanism can first automatically dispatches

Conclusions, insights, and future research

In this paper, we proposed a novel approach to the modeling and analysis of grid computing markets. In particular, we consider QoS, reliability, and risk management in a multiple criteria, multiple resource, multiple decision-maker and multiperiod grid computing network optimization framework. The proposed framework which is based on network equilibrium theory, allows for explicit modeling of decentralized decision-making behaviors, the maximization of net profit, and the minimization of risk,

Jose M. Cruz is an Assistant Professor of Operations and Information Management in the School of Business, University of Connecticut. His current research interests include complex decision-making on network systems (international financial networks, global supply chain networks, social and knowledge networks, and grid computing networks), corporate Social responsibility and risk management. His research has appeared in European Journal of Operational Research, Naval Research Logistics,

References (57)

  • R. Bapna et al.

    A market design for grid computing

    INFORMS Journal on Computing

    (2008)
  • F. Berman et al.

    The grads project: software support for high-level grid application development

    International Journal of High-Performance Computing Applications

    (2001)
  • R. Buyya

    Biogrid. virtual lab tools for data intensive computing on grid: drug design case study

  • R. Buyya, Economic-based Distributed Resource Management and Scheduling for Grid Computing, Ph.D. thesis, Monash...
  • R. Buyya et al.

    Economic models for resource management and scheduling in grid computing

    Concurrency and Computation: Practice and Experience

    (2002)
  • R. Buyya et al.

    Scheduling parameter sweep applications on global grids: a deadline and budget constrained cost-time optimization algorithm

    Software-Practice and Experiences

    (2005)
  • R. Buyya, D. Abramson, J. Giddy, Nimrod/g: An architecture for a resource management and scheduling system in a global...
  • P. Coddington

    Discworld, virtual data grids and grid applications

  • J. Cruz et al.

    Financial engineering of the integration of global supply chain networks and social networks with risk management

    Naval Research Logistics

    (2006)
  • A. Dogan et al.

    Scheduling independent tasks with qos requirements in grid computing with time-varying resource prices

  • C.L. Dumitrescu et al.

    Gruber: a grid resource usage sla broker

  • C. Ernemann

    Economic scheduling in grid computing

  • F. Facchinei et al.

    Finite-Dimensional Variational Inequalities and Complementarity Problems

    (2003)
  • I. Foster et al.

    Cloud computing and grid computing 360-degree compared

  • Gartner, Gartner says saas revenue within the enterprise application software market to total $9.2 billion in 2010

  • Globus.org, Globus alliance

  • S. Grid

    Sun grid computing utility. sun microsystems

  • S. Hwang et al.

    Grid workflow: a flexible failure handling framework for the grid

    Journal of Grid Computing

    (2003)
  • Cited by (8)

    • A bi-level decision support system for uncertain network design with equilibrium flow

      2015, Decision Support Systems
      Citation Excerpt :

      For instance, a spatial decision support system (SDSS) is developed in [5] to widely explore and examine the effects of different networked disruption scenarios. The proposed SDSS is helpful for decision makers in conveniently identifying critical network components and facilitate decisions about maintaining and enhancing network survivability. [7] also developed a network equilibrium model accounting for multi-criteria decision making behavior of various market participants for optimal pricing and resource allocation in a computational grid network.

    • Combining QoS prediction and customer satisfaction estimation to solve cloud service trustworthiness evaluation problems

      2014, Knowledge-Based Systems
      Citation Excerpt :

      We conduct a simulation experiment to demonstrate the effectiveness of our method. Based on the previous investigations [34–36], the commonly used cloud service trustworthy attributes include: Availability, Performance, Security, Privacy, Maturity, and Controllability. Their evaluation characteristics and weights are summarized as Table 4.

    • Network reliability based decision of Internet with multiple sources and multiple sinks

      2013, Decision Support Systems
      Citation Excerpt :

      The quality of service (QoS) is a critical task when seeking to realize the performance of an actual network system, and is a good tool to help the decision makers to improve the system. The subject of the QoS [4,8,25] of systems has been extensively studied for the past few decades. The QoS indicates the ability of the network to provide a consistent and predictable data transfer service that satisfies customers' applications, and a network reliability analysis is essential to maximize the use of network resources.

    View all citing articles on Scopus

    Jose M. Cruz is an Assistant Professor of Operations and Information Management in the School of Business, University of Connecticut. His current research interests include complex decision-making on network systems (international financial networks, global supply chain networks, social and knowledge networks, and grid computing networks), corporate Social responsibility and risk management. His research has appeared in European Journal of Operational Research, Naval Research Logistics, Environment & Planning B, International Journal of Production Research, International Journal of Production Economics, Decision Support Systems, Mathematical and Computer Modelling, Computational Management Science, Quantitative Finance, and other journals and book chapters.

    Zugang (Leo) Liu is an Assistant Professor of Business Administration in the Pennsylvania State University — Hazleton. His main research interest concerns decision making on complex network systems including global supply chain networks, energy and power networks, financial networks, transportation networks, and grid computing networks. His research has appeared in Naval Research Logistics, Journal of Banking and Finance, Transportation Research D, Transportation Research E, Mathematical and Computer Modelling, Computational Management Science, and other journals and book chapters.

    View full text