Virtual machine placement for elastic infrastructures in overbooked cloud computing datacenters under uncertainty

https://doi.org/10.1016/j.future.2017.09.021Get rights and content

Highlights

  • A first proposal for a complex IaaS environment for VMP problems considering service elasticity, including both vertical and horizontal scaling of cloud services, as well as overbooking of physical resources, including server (CPU and RAM) as well as networking resources (Ortigoza et al., 2016).

  • A two-phase optimization scheme for VMP problems, combining advantages of both online and offline VMP formulations in the proposed IaaS environment, introducing a prediction-based VMPr Triggering method to decide when to trigger a placement reconfiguration (Research Question 1) as well as an update-based VMPr Recovering method to decide what to do with VMs requested during placement recalculation times (Research Question 2).

  • A first scenario-based uncertainty approach for modeling the following relevant uncertain parameters of the proposed complex IaaS environment: (i) virtual resources capacities (vertical elasticity), (ii) number of VMs that compose cloud services (horizontal elasticity), (iii) utilization of CPU and RAM memory virtual resources (relevant for overbooking) and (iv) utilization of networking virtual resources (also relevant for overbooking).

  • A first formulation of a VMP problem considering the above mentioned contributions, for the optimization of the following four objective functions: (i) power consumption, (ii) economical revenue, (iii) resource utilization, as well as (iv) placement reconfiguration time.

  • An experimental evaluation of the presented two-phase optimization scheme against state-of-the-art alternatives for VMP problems, considering 400 different scenarios.

Abstract

Infrastructure as a Service (IaaS) providers must support requests for virtual resources in highly dynamic cloud computing environments. Due to the randomness of customer requests, Virtual Machine Placement (VMP) problems should be formulated under uncertainty. This work presents a novel two-phase optimization scheme for the resolution of VMP problems for cloud computing under uncertainty of several relevant parameters, combining advantages of online and offline formulations in dynamic environments considering service elasticity and overbooking of physical resources. In this context, a formulation of a VMP problem is presented, considering the optimization of the following four objective functions: (i) power consumption, (ii) economical revenue, (iii) resource utilization and (iv) reconfiguration time. The proposed two-phase optimization scheme includes novel methods to decide when to trigger a placement reconfiguration through migration of virtual machines (VMs) between physical machines (PMs) and what to do with VMs requested during the placement recalculation time. An experimental evaluation against state-of-the-art alternative approaches for VMP problems was performed considering 400 scenarios. Experimental results indicate that the proposed methods outperform other evaluated alternatives, improving the quality of solutions in a scenario-based uncertainty model considering the following evaluation criteria: (i) average, (ii) maximum and (iii) minimum objective function costs.

Introduction

Achieving an efficient resource management in cloud computing datacenters presents several research challenges, including relevant topics in resource allocation [1]. This work focuses on one of the most studied problems for resource allocation in cloud computing datacenters: the process of selecting which requested virtual machines (VMs) should be hosted at each available physical machine (PM) of a cloud computing infrastructure, commonly known as Virtual Machine Placement (VMP). This work proposes a complex Infrastructure as a Service (IaaS) environment for VMP problems, considering both service elasticity [2] and overbooking of physical resources [3].

To the best of the authors’ knowledge, there is no published work simultaneously taking into account elasticity and overbooking, directly related to the most relevant dynamic parameters in the literature on uncertain VMP problem considering multi-objective optimization. In order to model this complex IaaS environment for VMP problems, cloud services (i.e., inter-related VMs) are considered instead of isolated VMs [4].

It is worth remembering that VMP is a NP-Hard combinatorial optimization problem [5]. From an IaaS provider perspective, the VMP problem is mostly formulated as an online problem and must be solved with short time constraints [6].

Online decisions made along the operation of a dynamic cloud computing infrastructure negatively affects the quality of obtained solutions in VMP problems when comparing to offline decisions [7]. In this context, offline algorithms present a substantial advantage over online alternatives. Unfortunately, offline formulations are not appropriate for highly dynamic environments for real-world IaaS providers, where cloud services are requested dynamically according to current demand.

This work presents a two-phase optimization scheme, decomposing the VMP problem into two different sub-problems, combining advantages of online and offline VMP formulations considering a complex IaaS environment. The presented optimization scheme for the VMP problem introduces novel methods to decide when to trigger placement reconfigurations with migration of VMs between PMs (defined as VMPr Triggering) and what to do with cloud services requested during placement recalculation times (defined as VMPr Recovering).

For IaaS customers, cloud computing resources often appear to be unlimited and can be provisioned in any quantity at any required time [8]. Consequently, this work considers a basic federated-cloud deployment architecture for the VMP problem.

It is important to consider that more than 60 different objective functions have been proposed for VMP problems [6]. In this context, the number of considered objective functions may rapidly increase once a complete understanding of the VMP problem is accomplished for practical problems, where several different parameters should be ideally taken into account. Consequently, a renewed formulation of the VMP problem is presented, considering the optimization of the following four objective functions: (i) power consumption, (ii) economical revenue, (iii) resource utilization and (iv) reconfiguration time.

Due to the randomness of customer requests, VMP problems should be formulated under uncertainty [9]. This work presents a scenario-based uncertainty approach for modeling uncertain parameters, considering a two-phase optimization scheme for VMP problems in the proposed complex IaaS environments.

An experimental evaluation against state-of-the-art alternative approaches for VMP problems was performed considering 80 different workloads in 5 different CPU load scenarios, totalizing 400 experimental scenarios. Experimental results indicate that the proposed VMPr Triggering and Recovering methods of the presented two-phase optimization scheme outperform other evaluated alternatives, improving the quality of solutions.

In summary, the main contributions of this paper are:

  • A first proposal for a complex IaaS environment for VMP problems considering service elasticity, including both vertical and horizontal scaling of cloud services, as well as overbooking of physical resources, including server (CPU and RAM) as well as networking resources [4].

  • A two-phase optimization scheme for VMP problems, combining advantages of both online and offline VMP formulations in the proposed IaaS environment, introducing a prediction-based VMPr Triggering method to decide when to trigger a placement reconfiguration (Research Question 1) as well as an update-based VMPr Recovering method to decide what to do with VMs requested during placement recalculation times (Research Question 2).

  • A first scenario-based uncertainty approach for modeling the following relevant uncertain parameters of the proposed complex IaaS environment: (i) virtual resources capacities (vertical elasticity), (ii) number of VMs that compose cloud services (horizontal elasticity), (iii) utilization of CPU and RAM memory virtual resources (relevant for overbooking) and (iv) utilization of networking virtual resources (also relevant for overbooking).

  • A first formulation of a VMP problem considering the above mentioned contributions, for the optimization of the following four objective functions: (i) power consumption, (ii) economical revenue, (iii) resource utilization, as well as (iv) placement reconfiguration time.

  • An experimental evaluation of the presented two-phase optimization scheme against state-of-the-art alternatives for VMP problems, considering 400 different scenarios.

The remainder of this paper is structured in the following way: preliminary concepts and research challenges addressed in this work are introduced in Section 2, while related works and motivation of this work are summarized in Section 3. Section 4 presents the proposed uncertain VMP problem formulation considering four objectives, while Section 5 presents details on the design and implementation of evaluated alternatives to solve the proposed renewed formulation of the VMP problem. Experimental results are summarized in Section 6. Finally, conclusions and future work are left to Section 7.

Section snippets

Preliminary concepts and research challenges

The following sub-sections introduce relevant concepts related to the considered IaaS environments for VMP problems, a brief motivation for decomposing the VMP problem into two different sub-problems in a two-phase optimization scheme as well as uncertainty issues related to resource allocation in cloud computing. Additionally, the main challenges and research questions addressed in this work are also briefly introduced.

Related works and motivation

Chaisiri et al. studied in [27], [28] broker-oriented VMP problems under future demand and price uncertainty. To the best of the authors’ knowledge, there is no published work considering uncertainty of parameters for provider-oriented VMP problem formulations. Consequently, the following related works mainly focus on describing considered IaaS environments that proposed the utilization of two-phase optimization schemes for the VMP problem, as well as already proposed VMPr Triggering and VMPr

Uncertain VMP formulation

This section presents a formulation of the VMP problem under uncertainty considering a two-phase scheme for the optimization of the following objective functions: (i) power consumption, (ii) economical revenue, (iii) resource utilization and (iv) placement reconfiguration time. According to the taxonomy presented in [6], this work focuses on a provider-oriented VMP for federated-cloud deployments, considering a combination of two types of formulations: (i) online (i.e., iVMP) and (ii) offline

Evaluated algorithms

Taking into account that this work presents a novel uncertain VMP formulation considering a complex IaaS environment (see Section 4), there are no published alternatives to which we can compare the proposed algorithm. Therefore, the main goal of the experimental evaluation to be presented in Section 6 is to validate that the proposed VMPr Triggering and VMPr Recovering methods improve the quality of solutions, against adapted state-of-the-art alternatives that originally consider only partially

Experimental evaluation

The following sub-sections summarize the experimental environment as well as the main findings identified in the experiments performed as part of the simulations to validate the two-phase optimization scheme for VMP problems. The quality of solutions obtained by the evaluated algorithms in a scenario-based uncertainty model with 400 different scenarios was compared mainly considering the following evaluation criteria among solutions: (i) average, (ii) maximum and (iii) minimum objective

Conclusions and future work

This work presented a complex IaaS environment for VMP problems considering service elasticity, including both vertical and horizontal scaling of cloud services, as well as overbooking of physical resources, including both server (CPU and RAM) and networking resources (see Section 4.1).

The proposed complex IaaS environment for VMP problems was studied in a two-phase optimization scheme, combining advantages of both online and offline VMP formulations, where a novel prediction-based VMPr

Fabio López-Pires Fabio López Pires received a degree in Informatics Engineering (2010), a M.Sc. in Networks and Data Communications (2014) and D.Sc. in Computer Science (2017) from National University of Asunción in Paraguay. Currently, he works as Head of Distributed Systems and Parallel Computing Group at the Itaipu Technological Park. His research interests mainly focused on Cloud Computing, Evolutionary Algorithms and Multi-Objective Optimization.

References (44)

  • SpeitkampB. et al.

    A mathematical programming approach for server consolidation problems in virtualized data centers

    IEEE Trans. Serv. Comput.

    (2010)
  • López-PiresF. et al.

    A virtual machine placement taxonomy

  • F. López-Pires, B. Barán, A. Amarilla, L. Benítez, R. Ferreira, S. Zalimben, An experimental comparison of algorithms...
  • MellP. et al.

    The nist definition of cloud computing

    Natl. Inst. Stand. Technol.

    (2009)
  • MannZ.Á.

    Allocation of virtual machines in cloud data centers –A survey of problem models and optimization algorithms

    ACM Comput. Surv.

    (2015)
  • LiK. et al.

    Elasticity-aware virtual machine placement for cloud datacenters

  • WangW. et al.

    An availability-aware virtual machine placement approach for dynamic scaling of cloud applications

  • AnandA. et al.

    Virtual machine placement optimization supporting performance slas

  • BeloglazovA. et al.

    Optimal online deterministic algorithms and adaptive heuristics for energy and performance efficient dynamic consolidation of virtual machines in cloud data centers

    Concurr. Comput. Prac. Exper.

    (2012)
  • López-PiresF. et al.

    Multi-objective virtual machine placement with service level agreement: A memetic algorithm approach

  • CalcavecchiaN.M. et al.

    VM placement strategies for cloud scenarios

  • YueW. et al.

    Dynamic placement of virtual machines with both deterministic and stochastic demands for green cloud computing

    Math. Probl. Eng.

    (2014)
  • Cited by (31)

    • gVMP: A multi-objective joint VM and vGPU placement heuristic for API remoting-based GPU virtualization and disaggregation in cloud data centers

      2023, Journal of Parallel and Distributed Computing
      Citation Excerpt :

      Finally, in Section 6 we discuss final remarks. The problem of selecting and placing virtual machines in cloud data centers has been studied before in the literature [1,3,26,29,30,35,40,41,43,48,53,60]. However, GPU-enabled VM placement is an emerging problem and there are a few works addressing this issue.

    • An energy-aware resource deployment algorithm for cloud data centers based on dynamic hybrid machine learning

      2021, Knowledge-Based Systems
      Citation Excerpt :

      The cloud data center deploys virtual machines requested by cloud users on physical machines through its scheduling algorithm and then migrates or shuts down virtual machines based on real-time conditions to reduce energy consumption. Academia and industry have performed some research on scheduling optimization and energy consumption reduction [1–5], but these algorithms basically ignore the following issues. First, to meet the needs of cloud users, cloud service providers usually provide virtual machines with multiple configurations from which cloud users can choose, but there is a fixed ratio between the number of CPU cores and the size of memory.

    • The effect of server energy proportionality on data center power oversubscription

      2020, Future Generation Computer Systems
      Citation Excerpt :

      However, the focus on safe power oversubscription of data centers has been relatively limited. The concept of oversubscribing or overbooking resources can be found within various aspects of a data center, for example, virtual machines oversubscribing physical servers [29] or servers oversubscribing network bandwidth [30], and seems only natural to be extended to oversubscribing the power infrastructure as it increases profit as well as resource utilization. Determining the optimum number of servers that can be deployed within a given power limit is non-trivial [1].

    • An Improved Dynamic Fault Tolerant Management Algorithm during VM migration in Cloud Data Center

      2019, Future Generation Computer Systems
      Citation Excerpt :

      This enhances the reliability in CDC. The reduction of migration time can be achieved only when the VM is going to migrate back to the server that holds the VM’s previous virtual disk images. [19] dealt the problems of IAAS service elasticity and overbooking of resources.

    View all citing articles on Scopus

    Fabio López-Pires Fabio López Pires received a degree in Informatics Engineering (2010), a M.Sc. in Networks and Data Communications (2014) and D.Sc. in Computer Science (2017) from National University of Asunción in Paraguay. Currently, he works as Head of Distributed Systems and Parallel Computing Group at the Itaipu Technological Park. His research interests mainly focused on Cloud Computing, Evolutionary Algorithms and Multi-Objective Optimization.

    Prof. Benjamín Barán received a degree in Electronic Engineering (1983) from National University of Asunción in Paraguay, a M.Sc. in Electrical and Computer Engineering (1987) at Northeastern University in U.S.A. and a Ph.D. degree in Computer Science (1993) at the Federal University of Rio de Janeiro in Brazil. With more than 3 decades of teaching and research experience at several universities, he received the Paraguayan Science Award in 1996 and the Pan-American Prize of Scientific Computing in 2012, among a dozen of international awards. He was President of the Latin American Center on Informatics Studies (CLEI) and Research Coordinator at the National Computing Center (CNC) of the National University of Asunción. Dr. Barán is President of CBA S.A. His research interests focused on Cloud Computing, Evolutionary Computation, Multi-Objective Optimization and Optical Networks.

    Leonardo Benítez is a student of Informatics Engineering at the National University of Asunción in Paraguay. Currently, he is working as independent Software Developer. His research interest focused on Multi-Objective Optimization, Cloud Computing, Software Engineering and Web Technologies.

    Saúl Zalimben received a degree in Informatics Engineering (2017) from the National University of Asunción in Paraguay, working as independent Software Developer. His research interests focused on Multi-Objective Optimization, Cloud Computing, Datacenters and Software Engineering.

    Augusto Amarilla is a student of Informatics Engineering at the National University of Asunción in Paraguay. Currently doing research as part of his final project to achieve an Informatics Engineering degree. He also works as a full-stack developer and team leader at Software Natura. His research interests focused on Cloud Computing, Evolutionary Algorithms and Multi-Objective Optimization.

    View full text