Dynamic Resource Provisioning Research Articles

Dynamic Request Redirection and Resource Provisioning for Cloud-Based Video Services under Heterogeneous Environment

Cloud computing provides a new opportunity for Video Service Providers (VSP) to running compute-intensive video applications in a cost effective manner. Under this paradigm, a VSP may rent virtual machines (VMs) from multiple geo-distributed datacenters that are close to video requestors to run their services. As user demands are difficult to predict and the prices of the VMs vary in different time and region, optimizing the number of VMs of each type rented from datacenters located in different regions in a given time frame becomes essential to achieve cost effectiveness for VSPs. Meanwhile, it is equally important to guarantee users’ Quality of Experience (QoE) with rented VMs. In this paper, we give a systematic method called Dy namical Re quest Redire c t i on and R e source Pro v isioning ( DYRECEIVE ) to address this problem. We formulate the problem as a stochastic optimization problem and design a Lyapunov optimization framework based online algorithm to solve it. Our method is able to minimize the long-term time average cost of renting cloud resources while maintaining the user QoE. Theoretical analysis shows that our online algorithm can produce a solution within an upper bound to the optimal solution achieved through offline computing. Extensive experiments shows that our method is adaptive to request pattern changes along time and outperforms existing algorithms.

A REVIEW ON ENGERY EFFICIENT STRATEGY FOR TASK ALLOCATION IN CLOUD ENVIRONMENT

Although cloud computing is now becoming more advanced and matured as many companies have released their own computing platforms to provide services to public, but the research on cloud computing is still in its infancy. Apart from many other challenges of cloud computing, efficient management of energy is one of the most challenging research issues. In this paper we review the existing algorithm of dynamic resource provisioning and allocation algorithms and holistically work to boost data center energy efficiency and performance. This particular paper purposes a) heterogeneous workload and its implication on data centers energy efficiency b) solving the problem of VM resource scheduling to cloud applications

An elastic intrusion detection system for software networks

Internal users are the main causes of anomalous and suspicious behaviors in a communication network. Even when traditional security middleboxes are present, internal attacks may lead the network to outages or to leakage of sensitive information. In this article, we propose BroFlow, an Intrusion Detection and Prevention System based on Bro traffic analyzer and on the global network view of the software-defined networks (SDN) which is provided by the OpenFlow. BroFlow main contributions are (i) dynamic and elastic resource provision of traffic-analyzing machines under demand; (ii) real-time detection of DoS attacks through simple algorithms implemented in a policy language for network events; (iii) immediate reaction to DoS attacks, dropping malicious flows close of their sources, and (iv) near-optimal placement of sensors through a proposed heuristic for strategically positioning sensors in the network infrastructure, which is shared by multi-tenants, with a minimum number of sensors. We developed a prototype of the proposed system, and we evaluated it in a virtual environment of the Future Internet Testbed with Security (FITS). An evaluation of the system under attack shows that BroFlow guarantees the forwarding of legitimate packets at the maximal link rate, reducing up to 90 % of the maximal network delay caused by the attack. BroFlow reaches 50 % of bandwidth gain when compared with conventional firewalls approaches, even when the attackers are legitimate tenants acting in collusion. In addition, the system reduces the sensors number, while keeping full coverage of network flows.

Workload Prediction Using ARIMA Model and Its Impact on Cloud Applications’ QoS

As companies shift from desktop applications to cloud-based software as a service (SaaS) applications deployed on public clouds, the competition for end-users by cloud providers offering similar services grows. In order to survive in such a competitive market, cloud-based companies must achieve good quality of service (QoS) for their users, or risk losing their customers to competitors. However, meeting the QoS with a cost-effective amount of resources is challenging because workloads experience variation over time. This problem can be solved with proactive dynamic provisioning of resources, which can estimate the future need of applications in terms of resources and allocate them in advance, releasing them once they are not required. In this paper, we present the realization of a cloud workload prediction module for SaaS providers based on the autoregressive integrated moving average (ARIMA) model. We introduce the prediction based on the ARIMA model and evaluate its accuracy of future workload prediction using real traces of requests to web servers. We also evaluate the impact of the achieved accuracy in terms of efficiency in resource utilization and QoS. Simulation results show that our model is able to achieve an average accuracy of up to 91 percent, which leads to efficiency in resource utilization with minimal impact on the QoS.

Towards Cloud-based Asynchronous Elasticity for Iterative HPC Applications

Elasticity is one of the key features of cloud computing. It allows applications to dynamically scale computing and storage resources, avoiding over- and under-provisioning. In high performance computing (HPC), initiatives are normally modeled to handle bag-of-tasks or key-value applications through a load balancer and a loosely-coupled set of virtual machine (VM) instances. In the joint-field of Message Passing Interface (MPI) and tightly-coupled HPC applications, we observe the need of rewriting source codes, previous knowledge of the application and/or stop-reconfigure-and-go approaches to address cloud elasticity. Besides, there are problems related to how profit this new feature in the HPC scope, since in MPI 2.0 applications the programmers need to handle communicators by themselves, and a sudden consolidation of a VM, together with a process, can compromise the entire execution. To address these issues, we propose a PaaS-based elasticity model, named AutoElastic. It acts as a middleware that allows iterative HPC applications to take advantage of dynamic resource provisioning of cloud infrastructures without any major modification. AutoElastic provides a new concept denoted here as asynchronous elasticity, i.e., it provides a framework to allow applications to either increase or decrease their computing resources without blocking the current execution. The feasibility of AutoElastic is demonstrated through a prototype that runs a CPU-bound numerical integration application on top of the OpenNebula middleware. The results showed the saving of about 3 min at each scaling out operations, emphasizing the contribution of the new concept on contexts where seconds are precious.

Operation analysis of massively multiplayer online games on unreliable resources

We investigate the use of a new Massively Multiplayer Online Gaming (MMOG) ecosystem consisting of end-users, game providers, game operators, and Cloud resource providers, for autonomous, self-adaptive hosting and operation of MMOGs on unreliable resources. For this purpose, we developed an MMOG simulator compliant with our ecosystem, supported by a dynamic resource provisioning and load balancing algorithm. Using our simulator in which we inject traces collected from a real-world MMOG and resource characteristics from 16 Cloud providers, we study the impact on the involved actors by considering different resource availability levels. We highlight the advantages of dynamic resource allocation over the static overprovisioning with respect to two types of metrics: QoS offered to the clients and financial profit of game providers and operators.

Optimal partitioning of a multicore server processor

Optimal partitioning of a multicore server processor in a cloud computing environment, i.e., optimal system (virtual server) configuration for some given types of applications is considered in this paper. Such optimization is important for dynamic resource provision and on-demand server customization in a cloud computing environment for certain specific types of applications, such that the overall system performance is optimized without exceeding certain energy consumption budget. A multicore server processor is treated as a group of queueing systems with multiple servers, i.e., M/M/m queueing systems. The system performance measures are the average task response time and the average power consumption. Two core speed and power consumption models are considered, namely, the idle-speed model and the constant-speed model. Three problems are formulated and solved, namely, optimal multicore server processor partitioning, optimal multicore server processor partitioning with power constraint, and optimal power allocation. All these problems are well-defined optimization problems. It is shown that although these problems are sophisticated, they can be solved by numerical algorithms. Numerical data are demonstrated for each problem.

Multi-agent based dynamic resource provisioning and monitoring for cloud computing systems infrastructure

The cloud computing paradigm provides a shared pool of resources and services with different models delivered to the customers through the Internet via an on-demand dynamically-scalable form charged using a pay-per-use model. The main problem we tackle in this paper is to optimize the resource provisioning task by shortening the completion time for the customers' tasks while minimizing the associated cost. This study presents the dynamic resources provisioning and monitoring (DRPM) system, a multi-agent system to manage the cloud provider's resources while taking into account the customers' quality of service requirements as determined by the service-level agreement (SLA). Moreover, DRPM includes a new virtual machine selection algorithm called the host fault detection algorithm. The proposed DRPM system is evaluated using the CloudSim tool. The results show that using the DRPM system increases resource utilization and decreases power consumption while avoiding SLA violations.

A non-cooperative game-theoretic framework for resource allocation in network virtualization

Network virtualization is a new technology that aims at allowing multiple virtual networks (VNs) to coexist in the same equipment and to hide the heterogeneity of network infrastructure. The critical issue for a given infrastructure provider (InP), is how to provide customized and on demand resources for multiple service providers (SPs) with different Quality of Service (QoS) requirements. The should also fairly distribute the network physical resources, such as bandwidth of each physical link, buffer spaces, and processing cycles at each node. In this paper, we propose a new framework based on game theory, for both link and node dynamic allocation between multiple infrastructure providers (InPs) and service providers (SPs). Our approach focuses on provisioning and managing the physical resources in a virtualized network infrastructure. We propose a two-stage approach based on non-cooperative games. The first one is the resource negotiation game where the SP requests link and node resources from multiple InPs. The InP may reject the SP's request when it can potentially cause network congestion. The second stage of the proposal concerns dynamic resource provisioning and consists of two non cooperative games; the node allocation game and the link allocation game. The objective of both games is to allocate physical resources for different isolated VNs that are sharing the same physical substrate network. In the node allocation game, the proportional share mechanism is used. Every SP assigns a weight and submits a bid to each physical node and thereafter it receives a share proportional to its bid. In the link allocation game we investigate the case when multiple SPs compete for a portion of the available physical network capacity. Simulation results show that the proposed approach achieves high resource utilization, improves the network performance, and fairly distributes the link and node resources between multiple SPs.

Virtual Machine Migration Triggering using Application Workload Prediction

Dynamic provisioning of physical resources to Virtual Machines (VMs) in virtualized environments can be achieved by (i) vertical scaling-adding/removing attached resources from existing virtual machine and (ii) horizontal scaling-adding a new virtual machine with additional resources. The live migration of virtual machines across different Physical Machines (PMs) is a vertical scaling technique which facilitates resource hot-spot mitigation, server consolidation, load balancing and system level maintenance. It takes significant amount of resources to iteratively copy memory pages. Hence during the migration there may be too much overload which can affect the performance of applications running on the VMs on the physical server. It is better to predict the future workload of applications running on physical server for early detection of overloads and trigger the migration at an appropriate point where sufficient number of resources are available for all the applications so that there will not be performance degradation. This paper presents an intelligent decision maker to trigger the migration by predicting the future workload and combining it with predicted performance parameters of migration process. Experimental results shows that migration is triggered at an appropriate point such that there are sufficient amount of resources available (15–20% more resources than high valued threshold method) and no application performance degradation exists as compared to properly chosen threshold method for triggering the migration. Prediction with support vector regression has got decent accuracy with MSE of 0.026. Also this system helps to improve resource utilization as compared to safer threshold value for triggering migration by removing unnecessary migrations.

Using IDS fitted Q to develop a real-time adaptive controller for dynamic resource provisioning in Cloud's virtualized environment

Reinforcement learning (RL) is a powerful solution to adaptive control when no explicit model exists for the system being controlled. To handle uncertainty along with the lack of explicit model for the Cloud's resource management systems, this paper utilizes continuous RL in order to provide an intelligent control scheme for dynamic resource provisioning in the spot market of the Cloud's computational resources. On the other hand, the spot market of computational resources inside Cloud is a real-time environment in which, from the RL point of view, the control task of dynamic resource provisioning requires defining continuous domains for (state, action) pairs. Commonly, function approximation is used in RL controllers to overcome continuous requirements of (state, action) pair remembrance and to provide estimates for unseen statuses. However, due to the computational complexities of approximation techniques like neural networks, RL is almost impractical for real-time applications. Thus, in this paper, Ink Drop Spread (IDS) modeling method, which is a solution to system modeling without dealing with heavy computational complexities, is used as the basis to develop an adaptive controller for dynamic resource provisioning in Cloud's virtualized environment. The performance of the proposed control mechanism is evaluated through measurement of job rejection rate and capacity waste. The results show that at the end of the training episodes, in 90 days, the controller learns to reduce job rejection rate down to 0% while capacity waste is optimized down to 11.9%.

Multi-Cloud Provisioning and Load Distribution for Three-Tier Applications

Cloud data centers are becoming the preferred deployment environment for a wide range of business applications because they provide many benefits compared to private in-house infrastructure. However, the traditional approach of using a single cloud has several limitations in terms of availability, avoiding vendor lock-in, and providing legislation-compliant services with suitable Quality of Experience (QoE) to users worldwide. One way for cloud clients to mitigate these issues is to use multiple clouds (i.e., a Multi-Cloud). In this article, we introduce an approach for deploying three-tier applications across multiple clouds in order to satisfy their key nonfunctional requirements. We propose adaptive, dynamic, and reactive resource provisioning and load distribution algorithms that heuristically optimize overall cost and response delays without violating essential legislative and regulatory requirements. Our simulation with realistic workload, network, and cloud characteristics shows that our method improves the state of the art in terms of availability, regulatory compliance, and QoE with acceptable sacrifice in cost and latency.

Thoth

Following the 'no one size fits all' philosophy, active research in big data platforms is focusing on creating an environment for multiple 'one-size' systems to co-exist and cooperate in the same cluster. Consequently, it has now become imperative to provide an integrated management solution that provides a database-centric view of the underlying multi-system environment. We outline the proposal of DBMS + , a database management platform over multiple 'one-size' systems. Our prototype implementation of DBMS + , called Thoth , adaptively chooses a best-fit system based on application requirements. In this demonstration, we propose to showcase Thoth DM , a data management framework for Thoth which consists of a data collection pipeline utility, data consolidation and dispatcher module, and a warehouse for storing this data. We further introduce the notion of apps ; an app is a utility that registers with Thoth DM and interfaces with its warehouse to provide core database management functionalities like dynamic provisioning of resources, designing a multi-system-aware optimizer, tuning of configuration parameters on each system, data storage, and layout schemes. We will demonstrate Thoth DM in action over Hive, Hadoop, Shark, Spark, and the Hadoop Distributed File System. This demonstration will focus on the following apps : (i) Dashboard for administration and control that will let the audience monitor and visualize a database-centric view of the multi-system cluster, and (ii) Data Layout Recommender app will allow searching for the optimal data layout in the multi-system setting.

Dynamic Resource Provisioning for Energy Efficiency in Wireless Access Networks: A Survey and an Outlook

Traditionally, energy efficiency aspects have been included in the wireless access network design space only in the context of power control aimed at interference mitigation and for the increase of the terminal battery lifetime. Energy consumption of network components has also, for a long time, not been considered an issue, neither in equipment design nor in network planning and management. However, in recent years, with the user demand increasing at nearly exponential pace and margins rapidly shrinking, concerns about energy efficiency have been raised, with the objective of reducing network operational costs (not to mention the environmental issues). Installing more energy-efficient hardware does not seem to fully solve the problem, since wireless access networks are almost invariably (over)provisioned with respect to the peak user demand. This means that efficient resource management schemes, which are capable of controlling how much of the network infrastructure is actually needed and which parts can be temporarily powered off to save energy, can be extremely effective and provide quite large cost reductions. Considering that most of the energy in wireless access networks is consumed in the radio part, dynamic provisioning of wireless access network resources is crucial to achieving energy-efficient operation. The consensus on this approach in the research community has been wide in the last few years, and a large number of solutions have been proposed. In this paper, we survey the most important proposals, considering the two most common wireless access technologies, namely, cellular and WLAN. The main features of the proposed solutions are analyzed and compared, with an outlook on their applicability in typical network scenarios that also include cooperation between both access technologies. Moreover, we provide an overview of the practical implementation aspects that must be addressed to achieve truly energy-efficient wireless access networks, including current standardization work, and trends in the development of energy-efficient hardware.

SLA-based negotiation and brokering of cloud resources

Cloud computing represents an opportunity for IT users to reduce costs and increase efficiency providing an alternative way of using IT services. Elastic provisioning plays an important role by giving the possibility to get the best resources configuration that satisfies the application requirements. Even if there are many cloud providers, with a rich and various offer of technological solutions, above all at IAAS, however there is neither support of SLA negotiation, nor of SLA management yet. One of the most important issue in such a context is the lack of negotiation interfaces and mechanisms by current providers for dynamic provisioning, which instead only make available a configuration form to submit the request for a specific resource among the available ones. In this paper, we propose the design of a tool for SLA-based dynamic provisioning of cloud resources at IAAS, that offers to the user negotiation and brokering facilities by integrating multiple models. We present a prototype implementation of our architecture using the mOSAIC framework.

CloudMon: a resource‐efficient IaaS cloud monitoring system based on networked intrusion detection system virtual appliances

SummaryThe networked intrusion detection system virtual appliance (NIDS‐VA), also known as virtualized NIDS, plays an important role in the protection and safeguard of IaaS cloud environments. However, it is nontrivial to guarantee both of the performance of NIDS‐VA and the resource efficiency of cloud applications because both are sharing computing resources in the same cloud environment. To overcome this challenge and trade‐off, we propose a novel system, named CloudMon, which enables dynamic resource provision and live placement for NIDS‐VAs in IaaS cloud environments. CloudMon provides two techniques to maintain high resource efficiency of IaaS cloud environments without degrading the performance of NIDS‐VAs and other virtual machines (VMs). The first technique is a virtual machine monitor based resource provision mechanism, which can minimize the resource usage of a NIDS‐VA with given performance guarantee. It uses a fuzzy model to characterize the complex relationship between performance and resource demands of a NIDS‐VA and develops an online fuzzy controller to adaptively control the resource allocation for NIDS‐VAs under varying network traffic. The second one is a global resource scheduling approach for optimizing the resource efficiency of the entire cloud environments. It leverages VM migration to dynamically place NIDS‐VAs and VMs. An online VM mapping algorithm is designed to maximize the resource utilization of the entire cloud environment. Our virtual machine monitor based resource provision mechanism has been evaluated by conducting comprehensive experiments based on Xen hypervisor and Snort NIDS in a real cloud environment. The results show that the proposed mechanism can allocate resources for a NIDS‐VA on demand while still satisfying its performance requirements. We also verify the effectiveness of our global resource scheduling approach by comparing it with two classic vector packing algorithms, and the results show that our approach improved the resource utilization of cloud environments and reduced the number of in‐use NIDS‐VAs and physical hosts. Copyright © 2013 John Wiley & Sons, Ltd.

Dynamic Resource Provisioning in Cloud based on Queuing Model

One of the main aim of cloud computing is to provide bigger data center that will carter the storage needs of end user. In a data centre, server clusters are used to provide the required processing capability to get acceptable response time for interactive applications. In this paper, an interactive system based on queuing model is presented in which the cloud customer (CC) initially establishes the session to access the resources. The proposed model uses banker’s algorithm for resource allocation due to which deadlock for resource allocation among various processes is not possible. Moreover, by putting restriction on number of login users, resources are not choked out even in case of heavy demand of resources. The concept of resource allocation matrix helps the cloud service provider to predict the resource requirements in advance. Since new sessions can be established and existing sessions can be terminated, the number of logon users can change over time. Resources are dynamically allocated according to the requirements of the user. The results obtained are accurate in terms of predicting the minimum number of processor nodes required to meet the performance goal of an interaction application.

Advanced wireless and optical technologies for small-cell mobile backhaul with dynamic software-defined management

This article discusses novel wireless and optical technologies to address the radical new challenges of small cell mobile backhaul (MBH). Specifically, we examine 60 GHz and 70-80 GHz millimeter-wave technologies for high-capacity last mile and pre-aggregation backhaul, and orthogonal frequency-division multiple access passive optical networks as the optical technology complement for enabling flexible cost-efficient hybrid backhaul coverage. Flexible high-capacity hybrid millimeter wave/optical MBH network operation is next verified via network simulations in the context of a demanding, urban small-cell backhaul application. Finally, a novel software defined networking tool called the backhaul resource manager is introduced for automated dynamic resource provisioning and capacity-aware path computation that improves fairness, network utilization and end-to-end user quality of experience. The introduction of the novel wireless, optical, and software-defined technologies thus has the potential to truly revolutionize the future MBH network.

A benefit-aware on-demand provisioning approach for multi-tier applications in cloud computing

Dynamic resource provisioning is a challenging technique to meet the service level agreement (SLA) requirements of multi-tier applications in virtualization-based cloud computing. Prior efforts have addressed this challenge based on either a cost-oblivious approach or a cost-aware approach. However, both approaches may suffer frequent SLA violations under flash crowd conditions. Because they ignore the benefit gained that a multi-tier application continuously guarantees the SLA in the new configuration. In this paper, we propose a benefit-aware approach with feedback control theory to solve this problem. Experimental results based on live workload traces show that our approach can reduce resource provisioning cost by as much as 30% compared with a costoblivious approach, and can effectively reduce SLA violations compared with a cost-aware approach.

Affinity-aware modeling of CPU usage with communicating virtual machines

Use of virtualization in Infrastructure as a Service (IaaS) environments provides benefits to both users and providers: users can make use of resources following a pay-per-use model and negotiate performance guarantees, whereas providers can provide quick, scalable and hardware-fault tolerant service and also utilize resources efficiently and economically. With increased acceptance of virtualization-based systems, an important issue is that of virtual machine migration-enabled consolidation and dynamic resource provisioning. Effective resource provisioning can result in higher gains for users and providers alike. Most hosted applications (for example, web services) are multi-tiered and can benefit from their various tiers being hosted on different virtual machines. These mutually communicating virtual machines may get colocated on the same physical machine or placed on different machines, as part of consolidation and flexible provisioning strategies. In this work, we argue the need for network affinity-awareness in resource provisioning for virtual machines. First, we empirically quantify the change in CPU resource usage due to colocation or dispersion of communicating virtual machines for both Xen and KVM virtualization technologies. Next, we build models based on these empirical measurements to predict the change in CPU utilization when transitioning between colocated and dispersed placements. Due to the modeling process being independent of virtualization technology and specific applications, the resultant model is generic and application-agnostic. Via extensive experimentation, we evaluate the applicability of our models for synthetic and benchmark application workloads. We find that the models have high prediction accuracy — maximum prediction error within 2% absolute CPU usage.