Virtual Machine Provisioning Research Articles

Leveraging content similarity among VMI files to allocate virtual machines in cloud

To meet a myriad of customers’ demands, a large number of virtual machines (VMs) have to be provisioned simultaneously in cloud data centers. Provisioning is usually time consuming due to the large size of virtual machine image (VMI) file that needs to be transferred via networks. To address this issue, researchers attempt to leverage the content similarity among different VMI files to reduce the volume of transferred data. In the VM provisioning, the VM packing problem that minimizes the number of physical machines is another important issue. In this paper, our goal is to find a solution that tries to pack VMs to the minimum number of PMs as well as significantly reduces the total amount of transferred data. We formally define the problem of VM packing and minimizing the data transferring in the VM provisioning, named RTVD-VA. We first propose an approximation algorithm to minimize the amount of transferred VMI data when provisioning K VMs with the same size to a single physical machine. We then extend the algorithm to address the scenario of multiple PMs when using the minimum number of PMs. Based on the above two approximation algorithms, we propose a heuristic algorithm, namely Balance-Placement, to solve the problem in general cases. Our simulation results show that Balance-Placement outperforms existing solutions like PSO and Greedy-Cache and achieves the least amount of transferred data and the minimum number of used PMs in most scenarios.

Virtual machine provisioning, code management, and data movement design for the Fermilab HEPCloud Facility

The Fermilab HEPCloud Facility Project has as its goal to extend the current Fermilab facility interface to provide transparent access to disparate resources including commercial and community clouds, grid federations, and HPC centers. This facility enables experiments to perform the full spectrum of computing tasks, including data-intensive simulation and reconstruction. We have evaluated the use of the commercial cloud to provide elasticity to respond to peaks of demand without overprovisioning local resources. Full scale data-intensive workflows have been successfully completed on Amazon Web Services for two High Energy Physics Experiments, CMS and NOνA, at the scale of 58000 simultaneous cores. This paper describes the significant improvements that were made to the virtual machine provisioning system, code caching system, and data movement system to accomplish this work. The virtual image provisioning and contextualization service was extended to multiple AWS regions, and to support experiment-specific data configurations. A prototype Decision Engine was written to determine the optimal availability zone and instance type to run on, minimizing cost and job interruptions. We have deployed a scalable on-demand caching service to deliver code and database information to jobs running on the commercial cloud. It uses the frontiersquid server and CERN VM File System (CVMFS) clients on EC2 instances and utilizes various services provided by AWS to build the infrastructure (stack). We discuss the architecture and load testing benchmarks on the squid servers. We also describe various approaches that were evaluated to transport experimental data to and from the cloud, and the optimal solutions that were used for the bulk of the data transport. Finally, we summarize lessons learned from this scale test, and our future plans to expand and improve the Fermilab HEP Cloud Facility.

AUTO-SCALING AND ADJUSTMENT PLATFORM FOR CLOUD-BASED SYSTEMS

For customers of cloud-computing platforms it is important to minimize the infrastructure footprint and associated costs while providing required levels of Quality of Service (QoS) and Quality of Experience (QoE) dictated by the Service Level Agreement (SLA). To assist with that cloud service providers are offering: (1) horizontal resource scaling through provisioning and destruction of virtual machines and containers, (2) vertical scaling through changing the capacity of individual cloud nodes. Existing scaling solutions mostly concentrate on low-level metrics like CPU load and memory consumption which doesn’t always correlate with the level of SLA conformity. Such technical measures should be preprocessed and viewed from a higher level of abstraction. Application level metrics should also be considered when deciding upon scaling the cloud-based solution. Existing scaling platforms are mostly proprietary technologies owned by cloud service providers themselves or by third parties and offered as Software as a Service. Enterprise applications could span infrastructures of multiple public and private clouds, dictating that the auto-scaling solution should not be isolated inside a single cloud infrastructure.
 The goal of this paper is to address the challenges above by presenting the architecture of Auto-scaling and Adjustment Platform for Cloud-based Systems (ASAPCS). It is based on open-source technologies and supports integration of various low and high level performance metrics, providing higher levels of abstraction for design of scaling algorithms. ASAPCS can be used with any cloud service provider and guarantees that move from one cloud platform to another will not result in complete redesign of the scaling algorithm. ASAPCS itself is horizontally scalable and can process large amounts of real-time data which is particularly important for applications developed following the microservices architectural style. ASAPCS approaches the scaling problem in a nonstandard way by considering real-time adjustments of the application logic to be part of the scalability strategy if it can result in performance improvements.

Joint Optimization of Resource Provisioning in Cloud Computing

Cloud computing exploits virtualization to provision resources efficiently. Increasingly, Virtual Machines (VMs) have high bandwidth requirements; however, previous research does not fully address the challenge of both VM and bandwidth provisioning. To efficiently provision resources, a joint approach that combines VMs and bandwidth allocation is required. Furthermore, in practice, demand is uncertain. Service providers allow the reservation of resources. However, due to the dangers of over- and under-provisioning, we employ stochastic programming to account for this risk. To improve the efficiency of the stochastic optimization, we reduce the problem space with a scenario tree reduction algorithm, that significantly increases tractability, whilst remaining a good heuristic. Further we perform a sensitivity analysis that finds the tolerance of our solution to parameter changes. Based on historical demand data, we use a deterministic equivalent formulation to find that our solution is optimal and responds well to changes in parameter values. We also show that sensitivity analysis of prices can be useful for both users and providers in maximizing cost efficiency.

Feedback-Control & Queueing Theory-Based Resource Management for Streaming Applications

Recent advances in sensor technologies and instrumentation have led to an extraordinary growth of data sources and streaming applications. A wide variety of devices, from smart phones to dedicated sensors, have the capability of collecting and streaming large amounts of data at unprecedented rates. A number of distinct streaming data models have been proposed. Typical applications for this include smart cites & built environments for instance, where sensor-based infrastructures continue to increase in scale and variety. Understanding how such streaming content can be processed within some time threshold remains a non-trivial and important research topic. We investigate how a cloud-based computational infrastructure can autonomically respond to such streaming content, offering Quality of Service guarantees. We propose an autonomic controller (based on feedback control and queueing theory) to elastically provision virtual machines to meet performance targets associated with a particular data stream. Evaluation is carried out using a federated Cloud-based infrastructure (implemented using CometCloud)-where the allocation of new resources can be based on: (i) differences between sites, i.e., types of resources supported (e.g., GPU versus CPU only), (ii) cost of execution; (iii) failure rate and likely resilience, etc. In particular, we demonstrate how Little's Law-a widely used result in queuing theory-can be adapted to support dynamic control in the context of such resource provisioning.

Optimising the deployment of virtual machine image replicas in cloud storage clusters

With the immense proliferation of cloud-based applications, the diversity of cloud services exacerbates the complexity of virtual machine provisioning. Addressing the performance issue, a series of solutions have been proposed to optimise the VM replica deployment. However, the majority of the existing mechanisms concentrate on optimising the locations of VM replicas, or resizing the scales of different replica sets; without an overall consideration of both the platform and application features. The purpose of this paper is to optimise the deployment of VM image replicas in cloud storage clusters. We study the impacts of both application and platform features on the VM instance installation rate, and further propose a novel mechanism to optimise the deployment of VM image replicas. Our deployment mechanism has been implemented in a real cloud storage cluster. Experimental results show that our mechanism improves the overall throughput of a storage cluster and speeds up the image installation operations.

Prediction-based VM provisioning and admission control for multi-tier web applications

We present a prediction-based, cost-efficient Virtual Machine (VM) provisioning and admission control approach for multi-tier web applications. The proposed approach provides automatic deployment and scaling of multiple web applications on a given Infrastructure as a Service (IaaS) cloud. It monitors and uses collected resource utilization metrics itself and does not require a performance model of the applications or the infrastructure dynamics. The approach uses the OSGi component model to share VM resources among deployed applications, reducing the total number of required VMs. The proposed approach comprises three sub-approaches: a reactive VM provisioning approach called ARVUE, a hybrid reactive-proactive VM provisioning approach called Cost-efficient Resource Allocation for Multiple web applications with Proactive scaling (CRAMP), and a session-based adaptive admission control approach called adaptive Admission Control for Virtualized Application Servers (ACVAS). Performance under varying load conditions is guaranteed by automatic adjustment and tuning of the CRAMP and ACVAS parameters. The proposed approach is demonstrated in discrete-event simulations and is evaluated in a series of experiments involving synthetic as well as realistic load patterns.

Profit-Maximizing Virtual Machine Provisioning Based on Workload Prediction in Computing Cloud

Cloud providers now face the problem of estimating the amount of computing resources required to satisfy a future workload. In this paper, a virtual machine provisioning (VMP) mechanism is designed to adapt workload fluctuation. The arrival rate of forthcoming jobs is predicted for acquiring the proper service rate by adopting an exponential smoothing (ES) method. The proper service rate is estimated to guarantee the service level agreement (SLA) constraints by using a diffusion approximation statistical model. The VMP problem is formulated as a facility location problem. Furthermore, it is characterized as the maximization of submodular function subject to the matroid constraints. A greedy-based VMP algorithm is designed to obtain the optimal virtual machine provision pattern. Simulation results illustrate that the proposed mechanism could increase the average profit efficiently without incurring significant quality of service (QoS) violations.

Energy Efficient Live Virtual Machine Provisioning at Cloud Data Centers - A Comparative Study

Cloud computing offers utility-oriented IT services like: pervasive applications from consumer, scientific, and business domains based on a pay-as-you-go model. So, the workload in cloud environment is usually dynamic. At cloud data centers, different virtual machines (VMs) Provisioning techniques cause different CPU utilization. Therefore, VM Provisioning on PMs to improve resource utilization and reduce energy consumption is one of the major concerns for cloud providers. The problem of VM Provisioning includes queuing of VM requests, placing the VMs on hosts, and the optimization of the current VM allocation using Live Migration. The existing VM provisioning schemes are to optimize physical server and network resources utilization, but many of them also focus on optimizing multiple resources utilization simultaneously. The setting up of utilization thresholds for resources is one of the common optimization techniques. The ultimate aim of Cloud providers is to optimize resource usage and reduce energy consumption with the obligation of providing high Quality of Service (QoS) to customers, while maintaining the Service Level Agreements (SLAs). We surveyed various Live Virtual Machine Provisioning techniques and presented the comparison among few benchmark techniques based on adaptive utilization thresholds, as contribution to Green Cloud computing solutions. A performance evaluation study and comparison is done using the CloudSim toolkit. General Terms Cloud Computing, Cloud Provider, Energy Efficiency, SLA, Virtual Machine Allocation,

Large-scale virtual machines provisioning in clouds: challenges and approaches

The scale of global data center market has been explosive in recent years. As the market grows, the demand for fast provisioning of the virtual resources to support elastic, manageable, and economical computing over the cloud becomes high. Fast provisioning of large-scale virtual machines (VMs), in particular, is critical to guarantee quality of service (QoS). In this paper, we systematically review the existing VM provisioning schemes and classify them in three main categories. We discuss the features and research status of each category, and introduce two recent solutions, VMThunder and VMThunder+, both of which can provision hundreds of VMs in seconds.

Allocation of Virtual Machines in Cloud Data Centers—A Survey of Problem Models and Optimization Algorithms

Data centers in public, private, and hybrid cloud settings make it possible to provision virtual machines (VMs) with unprecedented flexibility. However, purchasing, operating, and maintaining the underlying physical resources incurs significant monetary costs and environmental impact. Therefore, cloud providers must optimize the use of physical resources by a careful allocation of VMs to hosts, continuously balancing between the conflicting requirements on performance and operational costs. In recent years, several algorithms have been proposed for this important optimization problem. Unfortunately, the proposed approaches are hardly comparable because of subtle differences in the used problem models. This article surveys the used problem formulations and optimization algorithms, highlighting their strengths and limitations, and pointing out areas that need further research.

Truthful Greedy Mechanisms for Dynamic Virtual Machine Provisioning and Allocation in Clouds

A major challenging problem for cloud providers is designing efficient mechanisms for virtual machine (VM) provisioning and allocation. Such mechanisms enable the cloud providers to effectively utilize their available resources and obtain higher profits. Recently, cloud providers have introduced auction-based models for VM provisioning and allocation which allow users to submit bids for their requested VMs. We formulate the dynamic VM provisioning and allocation problem for the auction-based model as an integer program considering multiple types of resources. We then design truthful greedy and optimal mechanisms for the problem such that the cloud provider provisions VMs based on the requests of the winning users and determines their payments. We show that the proposed mechanisms are truthful, that is, the users do not have incentives to manipulate the system by lying about their requested bundles of VM instances and their valuations. We perform extensive experiments using real workload traces in order to investigate the performance of the proposed mechanisms. Our proposed mechanisms achieve promising results in terms of revenue for the cloud provider.

EANAE: a Cloud Job Migration Based Fault Tolerance Services Using Energy-Aware Nash Auction Equilibrium Model

Cloud computing providers provision their resource into various kinds of Virtual Machine (VM) instances that are further allocated to the users for a particular time period. In current days, the major challenging task for cloud service providers is to build an efficient VM provisioning and allocation mechanisms. Several existing cloud provisioning mechanisms are based on the traditional cloud pricing policies and cost reservation techniques. These policies lacks the dynamic VM provisioning and cost reservation approaches. Many existing techniques also have the issues of job failure occurrence due to idle energy, complex computation of node prediction, and the network without consideration of success rate. To overwhelm the above existing limitations, the proposed strategy enhances the resource provisioning through the Energy Aware Nash Auction Equilibrium (EANAE) model. Subsequently, the optimal bidding mechanism is offered by the consideration of workload selection in terms of jobs deadline, average CPU time consumption, job submission, and job remaining time. Furthermore, the workflow optimization logics are presented by applying the migrations of jobs across multiple cloud VMs to optimize the cost. A cloud job migration based fault tolerant services have an effective scheduling mechanism, which in term maximize the revenue. The jobs are reallocated for available VM with the optimal cost and time. The experimental result exhibit better response time, migration frequency, execution time, memory utilization, and energy utilization than the existing method.

Design of Cost Function for VM Allocation in Cloud Computing

Cloud service, as one of major technologies in modern IT business, has attracted a lot of attentions from academia and industry. Due to the dynamics of user demands, service providers need VM(Virtual Machine) provisioning mechanism to estimate the amount of resources demanded by cloud users in the next time interval and to prepare the resources elastically. In this paper, we describe issues of VM provisioning and introduce the state-ofthe-art technologies for each issue. Besides, for the efficient VM provisioning, we propose a cost function to calculate the total expense of a service provider under workload fluctuation. To evaluate the effectiveness of our cost function, we show the performance evaluation with real workload data.

An integrated infrastructure in support of software development

This paper describes the design and the current state of implementation of an infrastructure made available to software developers within the Italian National Institute for Nuclear Physics (INFN) to support and facilitate their daily activity. The infrastructure integrates several tools, each providing a well-identified function: project management, version control system, continuous integration, dynamic provisioning of virtual machines, efficiency improvement, knowledge base. When applicable, access to the services is based on the INFN-wide Authentication and Authorization Infrastructure. The system is being installed and progressively made available to INFN users belonging to tens of sites and laboratories and will represent a solid foundation for the software development efforts of the many experiments and projects that see the involvement of the Institute. The infrastructure will be beneficial especially for small- and medium-size collaborations, which often cannot afford the resources, in particular in terms of know-how, needed to set up such services.

Managing Performance Overhead of Virtual Machines in Cloud Computing: A Survey, State of the Art, and Future Directions

Infrastructure-as-a-Service (IaaS) cloud computing offers customers (tenants) a scalable and economical way to provision virtual machines (VMs) on demand while charging them only for the leased computing resources by time. However, due to the VM contention on shared computing resources in datacenters, this new computing paradigm inevitably brings noticeable performance overhead (i.e., unpredictable performance) of VMs to tenants, which has become one of the primary issues of the IaaS cloud. Consequently, increasing efforts have recently been devoted to guaranteeing VM performance for tenants. In this survey, we review the state-of-the-art research on managing the performance overhead of VMs, and summarize them under diverse scenarios of the IaaS cloud, ranging from the single-server virtualization, a single mega datacenter, to multiple geodistributed datacenters. Specifically, we unveil the causes of VM performance overhead by illustrating representative scenarios, discuss the performance modeling methods with a particular focus on their accuracy and cost, and compare the overhead mitigation techniques by identifying their effectiveness and implementation complexity. With the obtained insights into the pros and cons of each existing solution, we further bring forth future research challenges pertinent to the modeling methods and mitigation techniques of VM performance overhead in the IaaS cloud.

Performance Enhancement for Network I/O Virtualization with Efficient Interrupt Coalescing and Virtual Receive-Side Scaling

Virtualization is a key technology in cloud computing; it can accommodate numerous guest VMs to provide transparent services, such as live migration, high availability, and rapid checkpointing. Cloud computing using virtualization allows workloads to be deployed and scaled quickly through the rapid provisioning of virtual machines on physical machines. However, I/O virtualization, particularly for networking, suffers from significant performance degradation in the presence of high-speed networking connections. In this paper, we first analyze performance challenges in network I/O virtualization and identify two problems—conventional network I/O virtualization suffers from excessive virtual interrupts to guest VMs, and the back-end driver does not efficiently use the computing resources of underlying multicore processors. To address these challenges, we propose optimization methods for enhancing the networking performance: 1) Efficient interrupt coalescing for network I/O virtualization and 2) virtual receive-side scaling to effectively leverage multicore processors. These methods are implemented and evaluated with extensive performance tests on a Xen virtualization platform. Our experimental results confirm that the proposed optimizations can significantly improve network I/O virtualization performance and effectively solve the performance challenges.

Guest Editors' introduction: Special section on optimizing the cloud

Cloud computing is defined as a pool of virtualized computer resources. Based on this virtualization, the cloud computing paradigm allows workloads to be deployed and scaled-out quickly through the rapid provisioning of virtual machines or physical machines. A cloud computing platform supports redundant, self-recovering, highly scalable programming models that allow workloads to recover from many inevitable hardware/software failures and monitoring resource use in real time for providing physical and virtual servers, on which the applications can run. A cloud computing platform is more than a collection of computer resources, because it provides a mechanism to manage those resources. In a cloud computing platform, software is migrating from the desktop into the "clouds" of the Internet, promising users anytime, anywhere access to their programs and data. The editor-in-chief provides an overview of the technical articles and features presented in this issue.

A Combinatorial Auction-Based Mechanism for Dynamic VM Provisioning and Allocation in Clouds

Cloud computing providers provision their resources into different types of virtual machine (VM) instances that are then allocated to the users for specific periods of time. The allocation of VM instances to users is usually determined through fixed-price allocation mechanisms that cannot guarantee an economically efficient allocation and the maximization of cloud provider's revenue. A better alternative would be to use combinatorial auction-based resource allocation mechanisms. This argument is supported by the economic theory; when the auction costs are low, as is the case in the context of cloud computing, auctions are especially efficient over the fixed-price markets because products are matched to customers having the highest valuation. The existing combinatorial auction-based VM allocation mechanisms do not take into account the user's demand when making provisioning decisions, that is, they assume that the VM instances are statically provisioned. We design an auction-based mechanism for dynamic VM provisioning and allocation that takes into account the user demand, when making provisioning decisions. We prove that our mechanism is truthful (i.e., a user maximizes its utility only by bidding its true valuation for the requested bundle of VMs). We evaluate the proposed mechanism by performing extensive simulation experiments using real workload traces. The experiments show that the proposed mechanism yields higher revenue for the cloud provider and improves the utilization of cloud resources.

Effective Technique for VM Provisioning on PaaS Cloud Computing Platform

In a Cloud Computing platform, Virtual Machine Provisioning involves instantiation of one or more Virtual Machines (VMs) that match the hardware and software requirements of an application and plays an important role in performance issues [4]. This paper discusses a VM Provisioning technique to reduce the response time in a cloud computing platform with PaaS type of deployment model when the physical servers hosting Virtual Machines in datacenters vary in their configurations. The method proposed in this paper is application specific and has been implemented on a test bed with VMware environment. Experimental results show that it can deliver better performance compared to random scheduling method for applications with repetitive computations