Conference PaperPDF Available

Order@Cloud: A VM Organisation Framework Based on Multi-Objectives Placement Ranking

Authors:

Abstract

This paper presents the implementation and tests of a flexible and extensible framework, named Order@Cloud, that improves the Virtual Machine placements of a Cloud. It receives new VMs on the Cloud and organises them by relocating their placements based on the Multiple-Objectives of the environment. These Objectives are represented by Rules, Qualifiers and Costs, which can be easily added, extended and prioritised. Based on Evolutionary and Greedy Searches, Order@Cloud theoretically guarantees the adoption of a better set of Placements. More specifically, it seeks the non-dominated solutions (Pareto Set) and compares them considering the implementation cost of the scenario and its benefits. In contrast to existing solutions, that address specific objectives, our framework was devised to be objective-agnostic and easily extensible, which enables the implementation of new and generic prioritised elements. To understand the applicability and performance of our solution we conducted experiments using a real Cloud environment and discuss its performance, flexibility and optimality.
Order@Cloud: A VM Organisation Framework
Based on Multi-Objectives Placement Ranking
Guilherme Arthur Geronimo
Federal University of Santa Catarina
Florian
´
opolis, Brazil
guilherme.geronimo@ufsc.br
Rafael Brundo Uriarte
IMT Institute For Advanced Studies
Lucca, Italy
rafael.uriarte@imtlucca.it
Carlos Becker Westphall
Federal University of Santa Catarina
Florian
´
opolis, Brazil
carlos.westphall@ufsc.br
Abstract—This paper proposes a flexible and extensible frame-
work, named Order@Cloud, to improve the Virtual Machine
Placement of a Cloud. It organises the Cloud VMs by relocating
them based on Multiple-Objectives of the environment. These Ob-
jectives are represented by Rules, Qualifiers, and Costs, which can
be easily added, extended and prioritised. Based on Evolutionary
Searches, Order@Cloud theoretically guarantees the adoption of
a better set of Placements. More specifically, it seeks the non-
dominated solutions (Pareto Set) and compare then considering
the implementation cost of the scenario and its benefits. In
contrast to existing solutions that address specific objectives,
our framework was devised to support any type of priority
and to be easily extensible, which enables the implementation
of new and generic prioritised elements. Moreover, we conducted
experiments using data from a real Cloud environments and show
the flexibility of our approach and its scalability.
KeywordsVirtual Machine Placement; Cloud Computing;
Multi-Objective Optimisation.
I. INTRODUCTION
Although Cloud Computing (CC) brings many benefits,
Cloud’s mismanagement usually accentuates problems related
to waste of resources. For example, VMs performance degrada-
tion due noisy neighbours [1], rise of thermal hotspots on data
center [1] and shortage of resources due constant migrations
[2]. Moreover, according to [3] approximately 30% of PMs
are actually idle, having an average usage ratio of 10%.
Many approaches were proposed to mitigate this problem,
such as Simulation-Based, Policy-Based, Bin Packing and Evo-
lutionary Algorithms (e.g. Ant Colony, Genetic Algorithms).
However, these approaches usually focus in specific objectives,
e.g. on decreasing the energy consumption, the number of Ser-
vice Level Agreement (SLA) violations and resource wastage.
To address this limitation, Xu et al. [4] propose a wider
definition for VMP (which is also used by [5], [6], [7]) as a
multiple objective (MO) optimisation problem that (simulta-
neously) tries to minimise the total resource wastage, power
consumption and thermal dissipation costs. This view extends
the previous approaches by modelling the VMP problem as
a MO optimisation issue, which also enables the managers to
consider other facets of VMP, for instance, internal policies
and Service Level Objectives (SLO).
Nevertheless, these many facets, which MOs cover, need to
be represent and implement in a standard manner. Since each
facet has its own nature, which could be quantifiable and/or
qualifiable, to the best of our knowledge, no suitable model
meet our needs to represent different types of evaluations at a
standard manner.
Moreover, to solve MOs conflicts and to guarantee fast
convergence to a good solution, proposals usually adopt Evo-
lutionary strategies with specific heuristics to build a set of
non-dominated scenarios (using Pareto). However, in these
solutions, the selection of the best scenario is another issue due
limited evaluation strategies and to considerable execution time
and cost. Moreover, none of them take into consideration, at
the same time, important characteristics of Clouds that impact
in this task, such as SLA, policies and scalability.
Considering these limitations, we propose an easily exten-
sible framework, named Order@Cloud, to address the VMP or-
ganisation problem. This framework adopts a flexible approach
that enables the assessment and comparison of from single
placements to the whole clusters, enabling evaluations with
grater precision. Moreover, it uses MO qualification functions
to select VMs which should be relocate, generates the possible
placement scenarios, filters the non-dominated results (Pareto
set) and selects the best result from this considering the costs
to reach the scenario and its benefits.
The Order@Cloud framework is easily extensible, takes
advantage of constraints and SLAs to reduce the computation
cost, enabling a fast local-optimal search. The main contribu-
tions of this framework are: (i) the support to generic multiple
objectives, (ii) objective prioritisation, (iii) generic environ-
ment rules and (iv) ability to compare different scenarios.
The rest of this paper is organized as follows: Section
II provides a background for the concepts and brings the
related works; Section III presents a Cloud model and details
the problem; Section IV presents the framework concepts,
goals and architecture; Section V provides implementation
guidelines to the Framework for Rules, Qualifiers and Cost
functions; Section VI contains the description and results
from the preliminary tests; Section VII concludes the article
addressing the future works and open issues.
II. BAC KG ROUND A ND RELATED WORKS
In the first part of this section we present some concepts
which help to understand our solution. Then, we analyse the
related works.
... As highlighted in the previous section, optimizing the VMP problem is foundational for delivering a reliable, quick, and efficient cloud gaming experience. In cloud computing, placement refers to the relationship between a VM and a physical machine (PM), or guest and host, respectively [90]. Effective VMP addresses latency issues, ensures efficient resource utilization, optimizes costs, and enhances QoS. ...
... This reduction in resource waste is particularly significant for match-based games that maintain high resource utilization throughout their operating hours. Furthermore, the study by Geronimo [90] proposed a framework named Order@ Cloud to tackle the VMP organization problem. Order@Cloud is a parallelizable, centralized, and recursive process that evaluates the current state of the cloud, including a list of VM placements, to improve the quality of these placements. ...
Article
Full-text available
Cloud computing is profoundly transforming the way IT services are implemented and provided to end-users. Gaming services are not exempt from this new trend. In fact, with cloud gaming or Gaming as a Service, players can enjoy a high-quality gaming experience even while using low-end devices with limited processing capabilities. However, achieving the delicate balance between gamers’ quality of experience and the provider’s net profit poses challenges in optimizing the cloud gaming experience. Implementing procedures and strategies to use the most suitable cloud servers and efficiently utilize their resources can help achieve such a balance. This entails allocating each player’s virtual machine to an appropriate physical server, simultaneously optimizing objectives like reducing resource waste, minimizing power consumption, and decreasing network transmission delays. In this article, we offer a thorough review of recent research on the Virtual Machine Placement (VMP) problem in the context of cloud gaming. We explore various facets, encompassing the architecture of cloud gaming, optimization methodologies, and cloud gaming services. Readers will acquire a comprehensive understanding of the challenges in cloud gaming research, with a specific focus on how optimizing the VMP problem can contribute to resolving associated issues.
... While the general aim remains the same, the Cloud model, the convergence method and the technical development have been significantly extended, in comparison to the previous version 5 . Among the novelties, we highlight the extension of the model to support multiple costs; new experiments; and a new method for provisioning and adaptation. ...
... We also quantify the scalability of Order@Cloud by measuring time to evaluate a real scenario and propose better placements with various number of VMs (350-600) and maximum number of migrations allowed (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). The results of the experiment show that even in the worst case scenario that considers 600 VMs and 20 migrations, it took approximately 62 seconds (1 in the Fig.8, the maximum value in the normalised time) to reach the best possible scenario using a standard hardware. ...
Article
Full-text available
We propose a flexible meta‐heuristic framework for virtual machine (VM) organisation, provisioning, and adaptation in the cloud domain, based on migration costs and environment constraints. Order@Cloud improves VM placements according to multiple objectives represented by rules, qualifiers, and improvement cost, which can be easily modified and extended. Order@Cloud theoretically guarantees the adoption of a better set of placements, after considering their costs and benefits, by prioritising the worst VM placements. While existing solutions address only specific objectives, our framework is objective‐agnostic and extensible, which enables the adoption and implementation of new policies and priorities. We conduct experiments using a real cloud environment data and discuss the framework's performance, flexibility, and optimality and provide insights on the challenges and benefits of deploying this framework.
... In the references [68], presented Order@Cloud, a flexible VMP framework that organizes the VMs by replacement of them based on the parameters like Rules, Qualifiers, and Costs. It can be easily added, extended and prioritized. ...
Article
Full-text available
Energy efficiency is one of the important issues in green cloud data centers (DCs). In this context, a virtual machine (VM) placement is one of the important techniques which can be used to achieve energy efficiency in such environments. Bio-inspired optimization algorithms are widely used in the literature to solve the VM placement (VMP) problem and different types of them are benefited to achieve energy efficiency while meeting Quality of Service (QoS) and user-specified constraints such as deadlines and cost. This paper presents a comprehensive survey and taxonomy of the bio-inspired VMP schemes. For this purpose, we first provide the essential concepts regarding the VMP and describe various objectives and factors which can be considered in this process. Then, we provide a taxonomy of VMP schemes regarding their applied optimization algorithms and compare their employed factors in the VMP process as well as simulator environments and the metrics which have been utilized in the verification of the investigated VMP frameworks. Finally, the concluding remarks and future researches directions are provided.
... A VM organization framework called order cloud was introduced in Geronimo et al 65 for relocating VMs using the four elements: rules, qualifiers, priorities, and costs. Rules define placement constraints while qualifiers assess the quality of placement like energy, load, and network, priorities aim to sort qualifiers based on weights and parameter, and costs define the objective functions. ...
Article
Full-text available
Cloud computing introduced a new paradigm in IT industry by providing on‐demand, elastic, ubiquitous computing resources for users. In a virtualized cloud data center, there are a large number of physical machines (PMs) hosting different types of virtual machines (VMs). Unfortunately, the cloud data centers do not fully utilize their computing resources and cause a considerable amount of energy waste that has a great operational cost and dramatic impact on the environment. Server consolidation is one of the techniques that provide efficient use of physical resources by reducing the number of active servers. Since VM placement plays an important role in server consolidation, one of the main challenges in cloud data centers is an efficient mapping of VMs to PMs. Multiobjective VM placement is generating considerable interest among researchers and academia. This paper aims to represent a detailed review of the recent state‐of‐the‐art multiobjective VM placement mechanisms using nature‐inspired metaheuristic algorithms in cloud environments. Also, it gives special attention to the parameters and approaches used for placing VMs into PMs. In the end, we will discuss and explore further works that can be done in this area of research. Reviewing VM consolidation problem plus the reasons behind using this technique while giving a deeper look into VM‐PM mapping subproblem. Classifying the current proposed multiobjective VM placement mechanisms based on their algorithmic approaches. Analyzing proposed mechanisms based on several factors like what objectives, resources, and the environment they have considered plus highlighting the main contributions and drawbacks of their approach. Finally, discussing the issues and limitations exist among these mechanisms and presenting the future works that can be done in this area.
Book
Full-text available
Os principais problemas associados à implementação e uso da gerência de redes e serviços ocorrem devido à grande quantidade de proposições, padrões e de diferentes produtos oferecidos no mercado, dificultando consideravelmente a tomada de decisão no que se refere a utilização da abordagem de gerência de redes e serviços mais adequada. Além disso, novas tendências na área de gerência de redes e serviços vêm sendo pesquisadas, entre estas destacam-se atualmente: gerência de redes sem fio, de sensores, óticas, futura internet, internet das coisas, internet espacial...; áreas funcionais de segurança, configuração, desempenho, contabilidade...; gerência de serviços de multimídia, data centers, grid, cloud, fog, edge virtualização...; e gerência centralizada, autonômica, distribuída, auto-gerência, baseada em políticas... Estas novas tendências vêm sendo pesquisadas no Laboratório de Redes e Gerência (LRG) da UFSC e a partir deste projeto as mesmas poderão ser aperfeiçoadas através das seguintes atividades deste projeto: A - Aperfeiçoamentos na Gerência Autonômica para Fog e IoT; B - Aperfeiçoamentos na Qualidade de Serviço para Aplicações de Tempo Real em IoT e Fog; C Aperfeiçoamentos na Segurança para Fog e IoT; D - Aperfeiçoamentos no Sistema de Resposta de Intrusão Autonômica em Cloud e IoT; E - Aperfeiçoamentos na Privacidade em Gerência de Identidade para Federações Dinâmicas em Cloud e IoT; e F - Aperfeiçoamentos no Controle de Acesso Dinâmico Baseado em Risco para uma Federação de Nuvem e IoT..
Chapter
The growth of power consumption in Cloud Computing systems is one of the current concerns of systems designers. In previous years, several studies have been carried out in order to find new techniques to decrease the cloud power consumption. These techniques range from decisions on locations for data centres to techniques that enable efficient resource management. Resource Allocation, as a process of Resource Management, assigns available resources throughout the data centre in an efficient manner, minimizing the power consumption and maximizing the system performance. The contribution presented in this chapter is an overview of the Resource Management and Resource Allocation techniques, which contribute to the reduction of energy consumption without compromising the cloud user and provider constraints. We will present key concepts regarding energy consumption optimization in cloud data centres. Moreover, two practical cases are presented to illustrate the theoretical concepts of Resource Allocation. Finally, we discuss the open challenges that Resource Management must face in the coming years.
Conference Paper
Full-text available
This paper proposes a flexible framework to improve the quality of Virtual Machine's placements, in Clouds. It organises them by relocating the VMs based on the Multiple-Objectives of the environment. These Objectives are represented by Rules, Qualifiers, and Costs, which can be extended and prioritised. Based on Evolutionary Searches, the framework theoretically guarantees the adoption of a better set of Placements. More specifically, it seeks the non-dominated solutions (Pareto's Dominance concept) and compares then considering the implementation cost of the scenario and its benefits. In contrast to existing solutions that address specific objectives, our framework was devised to support many types of objectives and to be easily extensible, which enables the implementation of new and generic prioritised elements. Moreover, we conducted experiments using data from a real Cloud environment and show the flexibility of our approach and its scalability.
Conference Paper
Full-text available
Managing and optimising cloud services is one of the main challenges faced by industry and academia. A possible solution is resorting to self-management, as fostered by autonomic computing. However, the abstraction layer provided by cloud computing obfuscates several details of the provided services, which, in turn, hinders the effectiveness of autonomic managers. Data-driven approaches, particularly those relying on service clustering based on machine learning techniques, can assist the autonomic management and support decisions concerning, for example, the scheduling and deployment of services. One aspect that complicates this approach is that the information provided by the monitoring contains both continuous (e.g. CPU load) and categorical (e.g. VM instance type) data. Current approaches treat this problem in a heuristic fashion. This paper, instead, proposes an approach, which uses all kinds of data and learns in a data-driven fashion the similarities and resource usage patterns among the services. In particular, we use an unsupervised formulation of the Random Forest algorithm to calculate similarities and provide them as input to a clustering algorithm. For the sake of efficiency and meeting the dynamism requirement of autonomic clouds, our methodology consists of two steps: (i) off-line clustering and (ii) on-line prediction. Using datasets from real-world clouds, we demonstrate the superiority of our solution with respect to others and validate the accuracy of the on-line prediction. Moreover, to show the applicability of our approach, we devise a service scheduler that uses the notion of similarity among services and evaluate it in a cloud test-bed.Managing and optimising cloud services is one of the main challenges faced by industry and academia. A possible solution is resorting to self-management, as fostered by autonomic computing. However, the abstraction layer provided by cloud computing obfuscates several details of the provided services, which, in turn, hinders the effectiveness of autonomic managers. Data-driven approaches, particularly those relying on service clustering based on machine learning techniques, can assist the autonomic management and support decisions concerning, for example, the scheduling and deployment of services. One aspect that complicates this approach is that the information provided by the monitoring contains both continuous (e.g. CPU load) and categorical (e.g. VM instance type) data. Current approaches treat this problem in a heuristic fashion. This paper, instead, proposes an approach, which uses all kinds of data and learns in a data-driven fashion the similarities and resource usage patterns among the services. In particular, we use an unsupervised formulation of the Random Forest algorithm to calculate similarities and provide them as input to a clustering algorithm. For the sake of efficiency and meeting the dynamism requirement of autonomic clouds, our methodology consists of two steps: (i) off-line clustering and (ii) on-line prediction. Using datasets from real-world clouds, we demonstrate the superiority of our solution with respect to others and validate the accuracy of the on-line prediction. Moreover, to show the applicability of our approach, we devise a service scheduler that uses the notion of similarity among services and evaluate it in a cloud test-bed.
Conference Paper
Full-text available
Cloud computing datacenters dynamically provide millions of virtual machines (VMs) in actual cloud markets. In this context, Virtual Machine Placement (VMP) is one of the most challenging problems in cloud infrastructure management, considering the large number of possible optimization criteria and different formulations that could be studied. VMP literature include relevant research topics such as energy efficiency, Service Level Agreement (SLA), Quality of Service (QoS), cloud service pricing schemes and carbon dioxide emissions; all of them with high economical and ecological impact. This work classifies an extensive up-to-date survey of the most relevant VMP literature proposing a novel taxonomy in order to identify research opportunities and define a general vision on this research area.
Conference Paper
Full-text available
The need of mechanisms to automate and regulate the interaction amongst the parties involved in the offered cloud services is exacerbated by the increasing number of providers and solutions that enable the cloud paradigm. This regulation needs to be defined through a contract, the so-called Service Level Agreement (SLA). We argue that the current solutions for SLA specification cannot cope with the distinctive characteristics of clouds. Therefore, in this paper we define a language, named SLAC, devised for specifying SLA for the cloud computing domain. The main differences with respect to the existing specification languages are: SLAC is domain specific, its semantics are formally defined in order to avoid ambiguity, it supports the main cloud deployment models, and it enables the specification of multi-party agreements. Moreover, SLAC supports the business aspects of the domain, such as pricing schemes, business actions and metrics. Furthermore, SLAC comes with an open-source software framework which enables the specification, evaluation and enforcement of SLAs for clouds. We illustrate potentialities and effectiveness of the SLAC language and its management framework by experimenting with an Open Nebula cloud system.
Conference Paper
Full-text available
This paper studies an evolutionary game theoretic framework for adaptive and stable application deployment in clouds. The framework, called Cielo, aids cloud operators to adapt the resource allocation to applications and their locations to the operational conditions in a cloud (e.g., workload and resource availability) with respect to multiple conflicting objectives (e.g., response time and power consumption). Moreover, Cielo theoretically guarantees that each application performs an evolutionarily stable deployment strategy, which is an equilibrium solution under given operational conditions. Simulation results verify this theoretical analysis, applications seek equilibria to perform adaptive and evolutionarily stable deployment strategies. Cielo outperforms well-known existing heuristics.
Article
In cloud computing, an important issue is virtual machine placement (VMP), selecting the most suitable set of physical hosts for a set of virtual machines. In this paper, we present a novel solution to the VMP problem called VMPMBBO. Our scheme treats the VMP problem as a complex system, and utilizes the biogeography-based optimization (BBO) technique to optimize the virtual machine placement that minimizes power consumption, resource waste, server unevenness, inter-VM traffic, storage traffic and migration time at the same time. Compared with three existing multi-objective VMP optimization algorithms, VMPMBBO has better convergence characteristics and is more computationally efficient. VMPMBBO is also robust. Extensive experiments are conducted using synthetic data from related literatures. The results confirm the effectiveness, efficiency, and robustness of the proposed approach. To the best of our knowledge, this work is the first approach that applies BBO and complex system optimization to virtual machine placement (VMP).
Article
he growing demand for performance observed in many organisations can still be considered the main motivator for the evolution of HPC (High Performance Computing) and, more recently, cloud environments. Part of this demand regards the need to deal with large and complex data sets, called big data. Performance improvement in such environments begins to be limited by energy consumption. Then, a clear understanding about the behaviour of applications and environments is essential to tackle this problem. Workload characterization is a well-known approach to describing and reproducing systems behaviour. However, there are several methodologies, techniques and parameters that can be considered for a workload characterization. As a result, we present a differentiated survey on workload characterization focusing on performance and energy efficiency improvement on HPC, cloud and big data environments. This survey aims to provide a scenario with a better understanding of state of the art in workload characterization in order to both improve performance and reduce energy consumption in HPC, cloud and big data environments. After an extensive review and classification of research works, our study indicates that around 56.4% of the papers reviewed offer contributions to performance and energy efficiency improvement. In addition, we observed a growing interest in this subject at a rate of 7.86% per year.
Conference Paper
Resource management in cloud platforms becomes an increasingly complex and daunting task surrounded by various challenges of stringent QoS requirements, service availability guaranteeing and escalating overhead of the infrastructure that resulted from operation costs and ecological impact. On the other hand, virtualization adds a greater flexibility to the resource managers in addressing such challenges. However, at the same time, it imposes a further challenge of added management complexity. Recently, we have proposed a resource management model for cloud platforms, which utilizes a new resource mapping formulation and relays on a hybrid virtualization framework in an attempt to realize a resource manager that intelligently adapts the available cloud resources to satisfy the conflicting objectives of the running applications and underlying infrastructures' requirements. Moreover, we have proposed state of the art Binary-Real coded Genetic Algorithm (BRGA), which has been applied successfully to a wide spectrum of global and constrained optimization problems from the known benchmark suites. In this paper, we aim to proceed by proposing a mathematical model and a modified version of BRGA to validate our model. In addition, we aim to evaluate the feasibility, effectiveness and scalability of our approach through simulation experiments.