Task-Based System Load Balancing in Cloud Computing Using Particle Swarm Optimization

Live virtual machine (VM) migration is a technique for achieving system load balancing in a cloud environment by transferring an active VM from one physical host to another. This technique has been proposed to reduce the downtime for migrating overloaded VMs, but it is still time- and cost-consuming, and a large amount of memory is involved in the migration process. To overcome these drawbacks, we propose a Task-based System Load Balancing method using Particle Swarm Optimization (TBSLB-PSO) that achieves system load balancing by only transferring extra tasks from an overloaded VM instead of migrating the entire overloaded VM. We also design an optimization model to migrate these extra tasks to the new host VMs by applying Particle Swarm Optimization (PSO). To evaluate the proposed method, we extend the cloud simulator (Cloudsim) package and use PSO as its task scheduling model. The simulation results show that the proposed TBSLB-PSO method significantly reduces the time taken for the load balancing process compared to traditional load balancing approaches. Furthermore, in our proposed approach the overloaded VMs will not be paused during the migration process, and there is no need to use the VM pre-copy process. Therefore, the TBSLB-PSO method will eliminate VM downtime and the risk of losing the last activity performed by a customer, and will increase the Quality of Service experienced by cloud customers.     

支持随机服务请求的云虚拟机按需物理资源分配方法

本文针对云平台按负载峰值需求配置处理机资源、提供单一的服务应用和资源需求动态变化导致资源利用率低下的问题,采用云虚拟机中心来同时提供多种服务应用.利用灰色波形预测算法对未来时间段内到达虚拟机的服务请求量进行预测,给出兼顾资源需求和服务优先等级的虚拟机服务效用函数,以最大化物理机的服务效用值为目标,为物理机内的各虚拟机动态配置物理资源.通过同类虚拟机间的全局负载均衡和多次物理机内各虚拟机的物理资源再分配,进一步增加服务请求量较大的相应类型的虚拟机的物理资源分配量.最后,给出了虚拟机中心基于灰色波形预测的按需资源分配算法ODRGWF.模拟实验表明所提算法能够有效提高云平台中处理机的资源利用率,对提高用户请求完成率以及服务质量都具有实际意义.     

基于LP&GR算法的多优先级虚拟机迁移策略研究*

如何对云计算中心的虚拟机(Virtual machine,VM)资源进行合理分配是近年来研究的一个热点问题。针对这一问题,本文提出了一种基于负载预测和灰色关联度(Load Prediction and Gray Relational,LP&GR)的虚拟机资源分配算法,通过预测虚拟机的负载状态防止虚拟机发生过载,并建立了基于虚拟机负载评价函数的决策分配模型。同时为虚拟机的迁移队列设置了多个优先级,结合了抢占式与非抢占式的执行策略,保证了虚拟机的有序迁移,并提高资源利用率。实验结果表明,结合多优先级的LP&GR算法同比其他算法能够有效实现云中心的负载均衡。     

基于动态调整阈值的虚拟机迁移算法

针对当前数据中心服务器能耗优化和虚拟机迁移时机合理性问题,提出一种基于动态调整阈值（DAT）的虚拟机迁移算法。该算法首先通过统计分析物理机历史负载数据动态地调整虚拟机迁移的阈值门限,然后通过延时触发和预测物理机的负载趋势确定虚拟机迁移时机。最后将该算法应用到实验室搭建的数据中心平台上进行实验验证,结果表明基于DAT的虚拟机迁移算法比静态阈值法关闭的物理机数量更多,云数据中心能耗更低。基于DAT的虚拟机迁移算法能根据物理机的负载变化动态迁移虚拟机,达到提高物理机资源利用率、降低数据中心能耗、提高虚拟机迁移效率的目的。     

An efficient job management of computing service using integrated idle VM resources for high-performance computing based on OpenStack

In recent years, various studies on OpenStack-based high-performance computing have been conducted. OpenStack combines off-the-shelf physical computing devices and creates a resource pool of logical computing. The configuration of the logical computing resource pool provides computing infrastructure according to the user’s request and can be applied to the infrastructure as a service (laaS), which is a cloud computing service model. The OpenStack-based cloud computing can provide various computing services for users using a virtual machine (VM). However, intensive computing service requests from a large number of users during large-scale computing jobs may delay the job execution. Moreover, idle VM resources may occur and computing resources are wasted if users do not employ the cloud computing resources. To resolve the computing job delay and waste of computing resources, a variety of studies are required including computing task allocation, job scheduling, utilization of idle VM resource, and improvements in overall job’s execution speed according to the increase in computing service requests. Thus, this paper proposes an efficient job management of computing service (EJM-CS) by which idle VM resources are utilized in OpenStack and user’s computing services are processed in a distributed manner. EJM-CS logically integrates idle VM resources, which have different performances, for computing services. EJM-CS improves resource wastes by utilizing idle VM resources. EJM-CS takes multiple computing services rather than single computing service into consideration. EJM-CS determines the job execution order considering workloads and waiting time according to job priority of computing service requester and computing service type, thereby providing improved performance of overall job execution when computing service requests increase.

     

虚拟计算环境下虚拟机资源负载均衡方法

针对虚拟机资源粒度大和迁移时传输数据量大的特点,提出一种基于虚拟机迁移的负载均衡方法。该方法利用负载阈值对宿主机后续时间节点的负载趋势进行预测,避免瞬时负载峰值触发的虚拟机迁移问题。在触发迁移后采用加权概率转发方式选择迁移目标节点,解决传统负载均衡技术中的群聚冲突问题。实验结果表明,在宿主机负载分布严重不平衡的情况下,该方法能有效改善系统性能。     

Virtual machine selection and placement for dynamic consolidation in Cloud computing environment

Dynamic consolidation of virtual machines (VMs) in a data center is an effective way to reduce the energy consumption and improve physical resource utilization. Determining which VMs should be migrated from an overloaded host directly influences the VM migration time and increases energy consumption for the whole data center, and can cause the service level of agreement (SLA), delivered by providers and users, to be violated. So when designing a VM selection policy, we not only consider CPU utilization, but also define a variable that represents the degree of resource satisfaction to select the VMs. In addition, we propose a novel VM placement policy that prefers placing a migratable VM on a host that has the minimum correlation coefficient. The bigger correlation coefficient a host has, the greater the influence will be on VMs located on that host after the migration. Using CloudSim, we run simulations whose results let draw us to conclude that the policies we propose in this paper perform better than existing policies in terms of energy consumption, VM migration time, and SLA violation percentage.     

基于虚拟机动态迁移的负载均衡策略

针对数据中心由于异构节点资源利用率不均衡导致的负载均衡问题,本文提出了一种基于动态阈值的迁移时机判决算法与基于负载类型感知的选择算法相结合的虚拟机动态迁移选择策略.该策略先通过监控全局负载度与高低负载节点占比动态调整状态阈值,并结合负载评估值判断迁移时机;再分析虚拟机负载类型,依据虚拟机与节点资源的依赖度、虚拟机当前内存带宽比和虚拟机贡献度选择待迁移虚拟机,并根据虚拟机与目的节点的资源匹配度与迁移代价选择目的节点,实现对高负载与低负载节点的虚拟机动态调整,从而优化节点资源配置问题.实验结果表明,该策略可以有效减少虚拟机迁移次数并保证数据中心服务质量,最终改善数据中心的负载均衡能力.     

虚拟机磁盘迁移技术研究与实现

已有基于内存的虚拟机迁移技术要求迁移源机和目标机之间必须共享网络磁盘,迁移性能受网络条件的影响很大,且在不支持"共享网络磁盘"的环境中,无法实现虚拟机迁移.针对上述问题,考虑到虚拟机磁盘作为虚拟机运行所需的持久化状态的封装体,所展现出的如高可用性、高效能、安全稳定等良好特性,提出基于磁盘的虚拟机迁移.通过设计虚拟磁盘驱动系统DiskMig,在实现虚拟机磁盘迁移的同时保证了其在源端和目的端的一致性.设计了"Transfer on Demand with Forward-ahead"(TOD&FA)算法快速同步源端和目的端磁盘文件差异.DiskMig采用自行设计的高效存储结构bitmap,使记录、查询虚拟机磁盘迁移过程中的大量I/O操作时仅需时间复杂度O(1).实验表明,DiskMig所采用的相关方法对虚拟机及其中应用程序运行性能的影响仅为5%左右,有效支持了虚拟机磁盘文件的迁移.     

虚拟机跨域迁移中的网络重定向模型

提出一个网络重定向模型,实现局域网或广域网环境下的虚拟机跨域迁移。利用地址解析协议同步策略,在源节点和迁移后的虚拟机之间快速建立单向IP隧道,将数据重定向至虚拟机。采用IP双栈方式,使虚拟机在保持原有会话的同时,通过新增IP响应所有新的连接请求。实验结果表明,该模型能减少部署时间,优化数据转发路径,降低响应延迟,实现跨域迁移。     

云数据中心网络中的先应式碎片整理算法

在云数据中心网络内,虚拟机（Virtual machine, VM）被动态创建和下线,这就导致资源碎片不被后续VM请求所利用。为解决上述问题,以最小化使用服务器数 量为目标的服务器整合技术被提出。虽然这种方法可以在某一时间段内减少资源碎片,但却付出了较大的VM迁移代价。因此本文提出了一种基于预测的先应式碎片 整理算法,在减少无效VM迁移的同时,将资源碎片重新整合为可用的连续资源,从而最大化VM收益。文中对此问题进行了数学定义,随后设计了启发式方法获取近似最优解。仿真结果表明,所提算法能够获取最大收益,并能够大幅度降低VM迁移成本。     

A historical probability based noise generation strategy for privacy protection in cloud computing

Cloud computing promises an open environment where customers can deploy IT services in pay-as-you-go fashion while saving huge capital investment in their own IT infrastructure. Due to the openness, various malicious service providers can exist. Such service providers may record service requests from a customer and then collectively deduce the customer private information. Therefore, customers need to take certain actions to protect their privacy. Obfuscation with noise injection, that mixes noise service requests with real customer service requests so that service providers will be confused about which requests are real ones, is an effective approach in this regard if those request occurrence probabilities are about the same. However, current obfuscation with noise injection uses random noise requests. Due to the randomness it needs a large number of noise requests to hide the real ones so that all of their occurrence probabilities are about the same, i.e. service providers would be confused. In pay-as-you-go cloud environment, a noise request will cost the same as a real request. Hence, with the same level of confusion, i.e. customer privacy protection, the number of noise requests should be kept as few as possible. Therefore in this paper we develop a novel historical probability based noise generation strategy. Our strategy generates noise requests based on their historical occurrence probability so that all requests including noise and real ones can reach about the same occurrence probability, and then service providers would not be able to distinguish in between. Our strategy can significantly reduce the number of noise requests over the random strategy, by more than 90% as demonstrated by simulation evaluation.     

求解复杂有时间窗装卸货问题的遗传算法

有时间窗装卸货问题是为一个车队安排最优的服务路径以满足客户的运输需求,每个客户的装卸货任务由一辆车完成,即在该客户的装货点装载一定数量的货物后运往该客户的卸货点,所有任务的完成必须满足车辆的容量约束、行程约束和客户装卸货点的时间窗约束。从多车库、多货物类型和满载三个方面对一般有时间窗装卸问题（PDPTW）进行了扩展,提出一种解决复杂PDPTW问题的遗传算法,实验结果表明,该算法能有效解决复杂PDPTW问题,并取得较好的优化结果。     

Admission control for media on demand services

Admission control software is used to make accept or deny decisions about incoming service requests to avoid overload. Existing media streaming software includes only limited support for admission control by allowing for predefined static rules. Such rules limit for example the number of requests that are allowed to enter the system during a certain time or define thresholds concerning the utilization level of a single resource such as network bandwidth. In media streaming applications, however, the bottleneck resource (CPU, Disk I/O, network bandwidth, etc.) might change over time depending on the current demand for different types of audio or video files. This paper proposes a model for adaptive admission control in the presence of multiple scarce resources. Opportunity costs for a service request are determined at the moment of an incoming request and compared to the revenue of a request in order to make an accept/deny decision. Opportunity costs are based on resource utilization, service resource requirements, expected future demand for services, and the revenue per accepted service. The model allows rejection of service requests early to reserve capacity required to perform future service requests with higher revenues. We describe a number of experiments to illustrate the benefits of adaptive admission control models over static admission control rules.     

Optimal virtual machine selection for anomaly detection using a swarm intelligence approach

Cloud computing plays a significant role in Healthcare Service (HCS) applications and rapidly improves it. A significant challenge is the selection of Virtual Machine (VM) in order to process a medical request. The optimal selection of VM increases the performance of HCS by minimizing the running time of the medical request and also substantially utilizes cloud resources. This paper presents a new idea for optimizing VM selection using a swarm intelligence approach called Analogous Particle swarm optimization (APSO) which works a cloud computing environment. To compute the running time of a medical request, three parameters are considered: Turnaround Time (TAT), Waiting time (WT), and CPU utilization. In addition, a selected optimal VM is used for predicting kidney disease. Early detection of kidney disease facilitates successful treatment. Here, the neural network is used as an automated technique to diagnose kidney disease. A set of experiments and comparisons were performed to analyze the proposed system (APSO and neural network). The results showed that the APSO model performed well, with an execution time of running all particle is 1 s (50 to 80%). Also, the proposed model improved the system efficiency by 5.6%. The precision of recognizing kidney disease using the neural network was 95.7% which outperfomed five other well-known classifiers.     

Towards an adaptive human-centric computing resource management framework based on resource prediction and multi-objective genetic algorithm

The complexity, scale and dynamic of data source in the human-centric computing bring great challenges to maintainers. It is problem to be solved that how to reduce manual intervention in large scale human-centric computing, such as cloud computing resource management so that system can automatically manage according to configuration strategies. To address the problem, a resource management framework based on resource prediction and multi-objective optimization genetic algorithm resource allocation (RPMGA-RMF) was proposed. It searches for optimal load cluster as training sample based on load similarity. The neural network (NN) algorithm was used to predict resource load. Meanwhile, the model also built virtual machine migration request in accordance with obtained predicted load value. The multi-objective genetic algorithm (GA) based on hybrid group encoding algorithm was introduced for virtual machine (VM) resource management, so as to provide optimal VM migration strategy, thus achieving adaptive optimization configuration management of resource. Experimental resource based on CloudSim platform shows that the RPMGA-RMF can decrease VM migration times while reduce physical node simultaneously. The system energy consumption can be reduced and load balancing can be achieved either.     

基于拍卖的数据中心资源匹配算法

云平台数据中心主机与负载均具有异构性,导致任务负载无法均衡利用主机各项资源。主机资源的非均衡利用最终造成总体资源利用率低,主机资源浪费,提高运营成本。针对云平台数据中心任务分配中各项资源无法均衡利用的问题,提出一种基于连续双向拍卖的虚拟机分配与迁移算法。该算法一方面利用多种启发式策略对数据中心主机和虚拟机进行筛选,将过载主机与欠载主机放入数据中心拍卖中;另一方面,构建买卖双方定价策略以及交易策略,形成完整的拍卖流程。同时,为解决多资源情况下的交易问题,提出基于资源匹配度的交易策略。仿真实验表明,文中方法通过引入资源匹配度,能够有效地匹配数据中心主机与虚拟机的各项资源,平衡各类资源利用率,提高整体资源利用率。     

Cost-based admission control for Internet Commerce QoS enhancement

In many e-commerce systems, preserving Quality of Service (QoS) is crucial to keep a competitive edge. Poor QoS translates into poor system resource utilisation, customer dissatisfaction and profit loss. In this paper, a cost-based admission control (CBAC) approach is described which is a novel approach to preserve QoS in Internet Commerce systems. CBAC is a dynamic mechanism which uses a congestion control technique to maintain QoS while the system is online. Rather than rejecting customer requests in a high-load situation, a discount-charge model which is sensitive to system current load and navigational structure is used to encourage customers to postpone their requests. A scheduling mechanism with load forecasting is used to schedule user requests in more lightly loaded time periods. Experimental results showed that the use of CBAC at high load achieves higher profit, better utilisation of system resources and service times competitive with those which are achievable during lightly loaded periods. Throughput is sustained at reasonable levels and request failure at high load is dramatically reduced.     

Video personalization in heterogeneous and resource-constrained environments

Access to multimedia data and multimedia services is becoming increasingly common in networked mobile environments. In such environments, both the mobile client devices and multimedia servers are typically resource constrained. Moreover, the mobile client population is often heterogeneous in terms of the clients’ preferences with regard to multimedia content, the clients’ quality of service requirements and system-level resource constraints. In order to provide a resource-constrained mobile client with its desired video content, it is often necessary to personalize the requested multimedia content in a manner that satisfies simultaneously the various client-specified content preferences and the system-level resource constraints. Also, in order to simultaneously reduce the client-experienced latency, provide optimal quality of service to the clients and ensure efficient utilization of server and network resources, it is necessary to perform client request aggregation on the server end. To this end, a video personalization strategy is proposed to provide mobile, resource-constrained clients with personalized video content that is most relevant to the clients’ requests while satisfying simultaneously multiple client-side system-level resource constraints. A client request aggregation strategy is also proposed to cluster client requests with similar video content preferences and similar client-side resource constraints such that the number of requests the server needs to process and service, and the client-experienced latency are both reduced simultaneously. The primary contributions of the paper are: (1) the formulation and implementation of a Multiple-choice Multi-dimensional Knapsack Problem (MMKP)-based video personalization strategy; and (2) the design and implementation of a client request aggregation strategy based on a multi-stage clustering algorithm. Experimental results comparing the proposed MMKP-based video personalization strategy to existing 0/1 Knapsack Problem (0/1KP)-based and the Fractional Knapsack Problem (FKP)-based video personalization strategies are presented. It is observed that: (1) the proposed MMKP-based personalization strategy includes more relevant video content in response to the client’s request compared to the existing 0/1KP-based and FKP-based personalization strategies; and (2) in contrast to the 0/1KP-based and FKP-based personalization strategies which can satisfy only a single client-side resource constraint at a time, the proposed MMKP-based personalization strategy is capable of satisfying multiple client-side resource constraints simultaneously. Experimental results comparing the client-experienced latency with and without the proposed server-side client request aggregation strategy are also presented. It is shown that the proposed client request aggregation strategy reduces the mean client-experienced latency without significant reduction in the average relevance of the delivered video content and without significant deviation in the client-side resources actually consumed by the delivered video content from the client-specified resource constraints.     

一种面向按需处理的空间数据服务模型与实验分析

通过传统空间数据服务系统获取的数据并不是满足用户需求的最终产品。对于地学领域的用户,通常需要根据特定应用目标,花费大量的时间进行数据预处理,从而满足特定的子区范围、波段和图像格式等。通过分析地学应用中针对数据源的操作要求,研究具有按需处理能力的空间数据服务模型,将各种分布的计算资源和数据资源联合起来,按照用于定制的需求,在为用户提供数据前,对数据进行一定的加工。通过对空间数据服务的基本操作进行分类和规范,研究建立兼容OGC WCS标准的数据服务机制,将大量基础性的数据预处理任务集成到数据获取流程中,以满足用户获取更高级别数据的需求。同时利用该模型在空间信息网格平台SIG(Spatial Information Grid)上进行验证,选择若干典型数据源,部署按需处理服务功能组件,对服务响应能力进行实验与评估。详细讨论了空间数据按需处理的流程,并将实验结果与传统方法进行对比分析。