Location-aware brokering for consumers in multi-cloud computing environments

https://doi.org/10.1016/j.jnca.2017.07.010Get rights and content

Abstract

The variety and complexity in cloud marketplaces is growing, making it difficult for cloud consumers to choose cloud services from multiple providers in an economic and suitable way by taking into account multiple objectives and constraints. In this paper, we present an extension of CloudSim implementing cloud management functionality to enable the assessment of consumer-oriented brokering schemes. The underlying discrete-event simulation framework allows evaluating their performance in more realistic operating conditions in a repeatable manner. We integrate brokering mechanisms to support a multi-criteria location-aware selection of virtual machines in multi-cloud environments by implementing a greedy heuristic and two large neighborhood search metaheuristics. Based on microbenchmarks of real cloud offerings and a diverse set of scenarios and workloads, we conduct simulation experiments to assess the performance of our approaches. The results show that approximately 10 – 12% of the total costs can be saved by using a large neighborhood search approach compared to the greedy heuristic. Finally, we analyze and discuss the trade-off between costs and latency as well as the impact of region constraints, showing, e.g., that latency improvements often come at a high price and a greater regional flexibility can lead to latency improvements while solely optimizing costs. Using real data of cloud marketplaces, we show that the proposed CloudSim extension can support decision makers as a tool for assessing cloud portfolios and market dynamics.

Introduction

The market of cloud services is rapidly growing and increasingly attracts new market participants offering and consuming cloud services (Cisco, 2015). This leads to a wide range of cloud providers and services entailing various features, pricing models, and service levels, standing in fierce competition to each other (Do et al., 2016). To maximize the utility of consumers using cloud services, costs and risks need to be minimized while business- and application-related requirements are met. As one cloud provider may not fulfill all requirements in the most economic way, the concept of having multiple clouds has been intensively discussed in the literature. In recent years, research focused on the federation of cloud providers collaborating and forming an inter-cloud to maintain all properties of the paradigm, including the impression of unlimited resources (see, e.g., Assis and Bittencourt, 2016, Toosi et al., 2014). Only some works have studied multi-clouds from a consumer perspective. That is, consumers simultaneously use cloud services of different cloud providers to further increase the business value. The main difference to federated clouds is that cloud providers not necessarily collaborate with each other and that the consumer is aware of using services from multiple providers. Although technological barriers are slowing down the development of multi-cloud applications in practice, current research approaches are promising (Petcu, 2013). As the decision making process is already difficult when adopting different cloud services from one cloud provider, multi-cloud environments will increase the need for brokers interacting with cloud providers on behalf of consumers to match consumer requirements with available cloud services. According to Gartner (2009), consumers depend on cloud brokers to unlock the potential of publicly available cloud services, for example, in terms of costs, quality, and flexibility.

In the area of cloud computing, one of the first definitions of a cloud broker is presented in Buyya et al. (2009). According to their definition, brokers mediate between consumers and cloud providers by purchasing and subleasing capacities to consumers. The role of a broker is not limited to a third-party organization or platform, but can also appear as a tool to coordinate and assign various resource requests within organizations. Classifications of cloud brokering schemes (e.g., proposed by Gartner in Plummer and Kenney, 2009) pay little attention to the optimization of cloud service discovery and selection notwithstanding the fact that it is one of the key elements to fully benefit from cloud marketplace offerings. A recent survey of existing broker solutions (Verginadis et al., 2014) identifies only a few tools with decision support capabilities, such as RightScale and DBCE. The business model of those tools is to solely provide a common platform for comparing different cloud service options (e.g., capacities, prices) based on individual consumer requirements. The same applies to real cloud marketplaces, which may provide tools to roughly estimate costs, but lack decision support for users aiming to select appropriate combinations and configurations of cloud services. Solving strategic and operational decision problems, however, typically involves multi-criteria objectives and requires efficient optimization techniques (Heilig and Voß, 2014).

In this paper, we consider a cloud brokering scheme aimed at optimizing the selection and utilization of virtual machine (VM) types offered in a cloud marketplace by multiple cloud providers. In this environment, brokers act on behalf of consumers intending to execute application and task requests in the cloud. The contributions of the paper are described as follows.

  • We extend the Cloud Service Purchasing Problem (CSPP), proposed in Heilig et al., 2016, to consider not only costs when assigning cloud resources, but also the network latency between the locations of consumers and cloud locations.

  • To simulate different brokering scenarios, we propose novel extensions of CloudSim. The extensions provide multi-cloud management functionality for embedding consumer-centric brokering schemes.

  • Given those extensions, we develop an optimization component for CloudSim to solve the extended CSPP by embedding two large neighborhood search metaheuristics.

  • Using real VM type descriptions and prices of three leading cloud providers and different workloads, we conduct a number of simulations for evaluating the performance of our approaches. The results reveal a competitive performance of the large neighborhood search approaches.

  • Finally, we demonstrate and discuss the effects of different decision making preferences towards costs and latency optimization and analyze the impacts of region constraints by slightly modifying the optimization problem.

The paper is structured as follows. In Section 2, we provide a brief overview on related work focusing on optimization approaches from a consumer perspective. The extended version of the tackled optimization problem is presented in Section 3. In Section 4, we briefly explain the extensions of CloudSim. The adapted large neighborhood search algorithms for solving the multi-criteria optimization problem are described in Section 5. Section 6 first describes the applied methodology for data collection as well as the simulation setup. Subsequently, the results of the conducted simulation experiments are presented and discussed. Finally, conclusions and directions for future research are presented in Section 7.

Section snippets

Related work

In recent years, only a few optimization approaches have been presented to facilitate the assignment and scheduling of tasks and applications in cloud environments from a consumer perspective. While earlier works focus on a single objective, namely cost optimization (e.g., Pandey et al., 2010, Van den Bossche et al., 2010, Chaisiri et al., 2012), recent works also aim to address performance aspects in the objective function (e.g., Lucas Simarro et al., 2011, Tordsson et al., 2012, Coutinho et

Extended cloud service purchasing problem

In this section, we explain the extended Cloud Service Purchasing Problem (CSPP) representing a multi-criteria optimization problem in the area of cloud computing. As such, it is an extension of the CSPP proposed in Heilig et al. (2016) to additionally consider the network performance between consumers and cloud DCs. Therefore, we refer to the mathematical model presented in Heilig et al. (2016) and focus on pointing out the main assumptions, constraints, and differences.

Generally, a broker is

Extension of CloudSim

As one of the major research goals is the support of consumer-centric brokering schemes in CloudSim, this section explains the implemented extensions for supporting the management of cloud resources in multi-cloud environments. In general, the CloudSim toolkit supports modeling and simulation of cloud computing environments and has been used for investigations by a large part of the research community (Calheiros et al., 2011). CloudSim contains a collection of packages and libraries written in

Optimization approaches

In this section, we explain the implemented large neighborhood search approaches designed for solving the extended CSPP in CloudSim. With respect to exact approaches, only small problem instances of the single objective CSPP could be solved to optimality (up to 10 requests) in the previous work of Heilig et al. (2016). For those instances, competitive results are achieved when applying the large neighborhood search algorithms, where the gap to the optimal solution is on average 2.88% and 3.85%

Performance evaluation

Using the implemented CloudSim extensions, we conduct multiple simulation experiments to evaluate the proposed algorithms and to explore the trade-off between costs and latency in multi-cloud environments. In the following, we first briefly describe the methodology for collecting and preparing input data as well as the configurations and scenarios that are applied in the different simulation experiments. Finally, we present the simulation results of the different experiments and discuss major

Conclusions and future directions

With the rapid adoption of cloud computing and the growing range of cloud marketplace offerings of different cloud providers, consumer-oriented brokering schemes supporting an economically viable selection and utilization of suitable cloud services have become essential. In this paper, we present novel extensions of CloudSim allowing to model, simulate, and evaluate brokering schemes that support cloud consumers in selecting and utilizing cloud resources in multi-cloud environments. Thus, the

Acknowledgments

We would like to thank Adel Nadjaran Toosi for his constructive suggestions on this article.

Leonard Heilig is a Ph.D. candidate at the Institute of Information Systems at the University of Hamburg. He holds a B.Sc. (University of Münster, Germany) and a M.Sc. (University of Hamburg, Germany) in Information Systems. His current research interest is centered around the adoption and utilization of cloud computing in contemporary enterprises with the focus on IT governance, decision support, mobile cloud applications, and its application in the maritime sector. He spent some time at the

References (32)

  • S. Chaisiri et al.

    Optimization of resource provisioning cost in cloud computing

    IEEE Trans. Serv. Comput.

    (2012)
  • C. Chekuri et al.

    On multi-dimensional packing problems

    SIAM J. Comput.

    (2012)
  • Cisco, 2015. Cisco Global Cloud Index: Forecast and Methodology, 2014–2019, URL...
  • Coutinho, R.d.C., Drummond, L.M.A., Frota, Y., 2013. Optimization of a cloud resource management problem from a...
  • Gartner, 2009. Gartner Says Cloud Consumers Need Brokerages to Unlock the Potential of Cloud Services. URL...
  • N. Grozev et al.

    Regulations and latency-aware load distribution of web applications in multi-clouds

    J. Supercomput.

    (2016)
  • Cited by (0)

    Leonard Heilig is a Ph.D. candidate at the Institute of Information Systems at the University of Hamburg. He holds a B.Sc. (University of Münster, Germany) and a M.Sc. (University of Hamburg, Germany) in Information Systems. His current research interest is centered around the adoption and utilization of cloud computing in contemporary enterprises with the focus on IT governance, decision support, mobile cloud applications, and its application in the maritime sector. He spent some time at the University of St Andrews (Scotland, UK) and, most recently, at the Cloud Computing and Distributed Systems (CLOUDS) Laboratory at the University of Melbourne, Australia. Currently he serves as guest editor for the Information Technology&Management journal special issue on information systems and big data in maritime logistics and seaports. His industry experiences in cloud computing include working as a software engineer at Adobe Systems participating in the development of their PaaS environment and at Fiducia&GAD IT, one of the largest IT providers in Germany, focusing on cloud-based banking applications. Further experiences include positions at Beiersdorf Shared Services and Airbus Group Innovations on security management.

    Dr. Rajkumar Buyya is a Fellow of IEEE, Professor of Computer Science and Software Engineering, Future Fellow of the Australian Research Council, and Director of the Cloud Computing and Distributed Systems (CLOUDS) Laboratory at the University of Melbourne, Australia. He is also serving as the founding CEO of Manjrasoft, a spin-off company of the University, commercializing its innovations in Cloud Computing. He has authored over 525 publications and seven text books including “Mastering Cloud Computing” published by McGraw Hill, China Machine Press, and Morgan Kaufmann for Indian, Chinese and international markets respectively. He also edited several books including “Cloud Computing: Principles and Paradigms” (Wiley Press, USA, Feb 2011). He is one of the highly cited authors in computer science and software engineering worldwide (h-index=106, g-index=221, 53,500+ citations). Recently, Dr. Buyya is recognized as “2016 Web of Science Highly Cited Researcher” by Thomson Reuters. Software technologies for Grid and Cloud computing developed under Dr. Buyya's leadership have gained rapid acceptance and are in use at several academic institutions and commercial enterprises in 40 countries around the world. Dr. Buyya has led the establishment and development of key community activities, including serving as foundation Chair of the IEEE Technical Committee on Scalable Computing and five IEEE/ACM conferences. These contributions and international research leadership of Dr. Buyya are recognized through the award of “2009 IEEE Medal for Excellence in Scalable Computing” from the IEEE Computer Society TCSC. Manjrasoft's Aneka Cloud technology developed under his leadership has received “2010 Frost&Sullivan New Product Innovation Award”. He served as the founding Editorin- Chief of the IEEE Transactions on Cloud Computing. He is currently serving as Co-Editor-in-Chief of Journal of Software: Practice and Experience, which was established over 45 years ago. For further information on Dr. Buyya, please visit his cyberhome: http://www.buyya.com

    Dr. Stefan Voß is professor and director of the Institute of Information Systems at the University of Hamburg. He also holds a visiting position at PUCV in Valparaiso, Chile. Previous positions include full professor and head of the department of Business Administration, Information Systems and Information Management at the University of Technology Braunschweig (Germany) from 1995 up to 2002. He holds degrees in Mathematics (diploma) and Economics from the University of Hamburg and a Ph.D. and the habilitation from the University of Technology Darmstadt. His current research interests are in quantitative/information systems approaches to supply chain management and logistics including public mass transit and telecommunications. He is author and co-author of several books and numerous papers in various journals. Stefan Voß serves on the editorial board of some journals including being Editor of Netnomics and Editor of Public Transport. He is frequently organizing workshops and conferences. Furthermore, he is consulting with several companies.

    View full text