MapReduce scheduler by characterizing performance interference

Virtual organization clusters are systems comprised of virtual machines that provide dedicated computing clusters for each individual virtual organization. The design of these clusters allows individual virtual machines to be independent of the underlying physical hardware, potentially allowing virtual clusters to span multiple grid sites. A major challenge in using virtual organization clusters as a grid computing abstraction arises from the need to schedule and provision physical resources to run the virtual machines. This paper describes a virtual cluster scheduler implementation based on the Condor high throughput computing system. By means of real-time monitoring of the Condor job queue, virtual machines that belong to individual virtual organizations are provisioned and booted. Jobs belonging to each virtual organization are then run on the organization-specific virtual machines, which form a cluster dedicated to the specific organization. Once the queued jobs have executed, the virtual machines are terminated, thereby allowing the physical resources to be re-claimed. Tests of this system were conducted using synthetic workloads, demonstrating that dynamic provisioning of virtual machines preserves system throughput for all but the shortest-running of grid jobs, without undue increase in scheduling latency.

In cloud computing, virtualization is accomplished by equipping with a virtual switch to connect between physical machine and virtual machine. The virtual switch which is usually written coded in software provides a flexible configuration and management. In this paper, we use kernel based virtual machine (KVM) to deploy the virtual server cluster and apply Open vSwitch for the virtualized network environment. Our platform provides service with single entry point so that the virtual server cluster is fully transparent to end users. The implementation of load balancing is carried out by using Open vSwitch with an Openflow controller. Openflow controller dynamically manages the flow table in Open vSwitch through secure channel by Openflow protocol. With the platform, we can also deploy a virtual cluster across two Local Area Networks (LANs) through Generic Routing Encapsulation tunnel protocol. Upon a request, the Open vSwitch forwards the traffic to the virtual server cluster upon the loading policy to achieve load balancing.

The quality of the scheduling method for virtualization clusters directly affects the performance of the entire cluster. Improper scheduling can even eliminate the advantages of virtualization clusters and reduce the performance of the entire system. Aiming at the problems of low resource utilization and poor performance caused by improper resource scheduling in cluster mode, a comprehensive heuristic resource scheduling method based on virtualization cluster - Resource Type Priority List Scheduling (RTPLS) algorithm is proposed. The entire algorithm includes two stages: resource monitoring and virtual machine scheduling. On the basis of considering multiple threshold triggering conditions such as CPU, memory, and network in the host load judgment during virtual machine scheduling, the host’s resource oscillation granularity and resource weight are introduced to comprehensively evaluate the resource load situation of the entire host, avoiding frequent migration of virtual machines caused by the oscillation of host resource load. In the selection of target virtual machine migration, in order to minimize the number of virtual machine migrations as much as possible, the compatibility coefficient λ of virtual machines is introduced, and the migration is sorted according to the priority list of λ to reduce the server load and energy consumption caused by virtual machine migration with the minimum number of virtual machine migrations. When selecting the target server, in order to minimize the load on the target virtual machine, target server relevance is introduced. Finally, by testing the algorithm in practical environments, it was found that the RTPLS algorithm can significantly improve the efficiency of host load balancing adjustments and enhance resource utilization.

As server consolidations based on the virtualization techniques become popular and cloud services continue to grow rapidly, more and more data centers are being built to accommodate a number of virtual clusters running various workloads. Since these virtual clusters often share the resources provided by physical machines (PMs), it is more likely that the interferences between virtual machines (VMs) affect the performance of applications running on top of the virtual clusters. While a lot of studies have proposed different virtual machine placement algorithms to investigate this issue, the problem caused by network performance variability still remains as a challenging issue. Since they usually ignore the CPU overhead to process the communications between VMs, the network bandwidth allocated to a VM cannot be fully utilized when a PM has not enough CPU resources to cover the CPU overhead for VM networking functions. This results in unpredictable application performance running on the virtual clusters. This paper proposes a virtual machine placement algorithm that considers the CPU overhead incurred to reserve network bandwidth in a virtualized data center environment. In order to decide the CPU overhead necessary to guarantee the network bandwidth allocated to a VM, a performance model based on standard linear regression using the data collected from a real environment is used. By comparing the amount of CPU resource available in the driver domain with the CPU overhead obtained from the performance model, the proposed algorithm decides whether the network bandwidth for the VM can be provided or not and selects an appropriate location for VM placement. The benchmarking results show that the proposed algorithm guarantees the network bandwidth allocated to each VM without violations when the CPU resources are shared by multiple VMs.

A virtualized cloud platform may host several cloud services/applications simultaneously. Each hosted cloud service usually runs on a virtual cluster, consisting of several virtual machines (VMs) with a virtual network. In many cases, the virtual clusters share the overall physical network bandwidth. Thus, the quality of the cloud services sharing the same physical networks can be seriously affected when one of the cloud services produces a high traffic load. In this paper, we propose a mechanism to solve the mentioned issue. The proposed mechanism utilizes a logically centralized QoS controller to allocate network bandwidth to virtual clusters dynamically. The QoS controller uses a monitor to continuously gather network usage of each VM, and calculates the bandwidth consumption in the granule of virtual clusters. A Rate Limiter is bundled to each physical machine for triggering bandwidth re-distribution. The proposed mechanism is implemented on the OpenStack-based cloud platform, SAMEVED, which is developed to support on-line network security experiments. We implemented several network bandwidth re-distribution strategies on the proposed mechanism. We emulate the real network flows in the platform, and the experiment results show that, the proposed approach can effectively reduce nearly one third of packet loss rate when running a video streaming service in a high network traffic scenario.

Recently, virtualization of hardware resources to run multiple instances of independent virtual machines over physical hosts has gained popularity due to an industry-wide focus on the need to reduce the cost of operation of an enterprise computing infrastructure. Xen is an open source hypervisor that provides a virtual machine abstraction layer which is very similar to the underlying physical machine. Using multiple physical hosts, each hosting multiple virtual machines over a VMM like Xen, system administrators can setup a high-availability virtual cluster to meet the ever-increasing demands of their data centers. In such an environment, the Xen hypervisor enables live migration of individual virtual machine instances from one physical node to another without significantly affecting the performance of the applications running on a target virtual machine. This paper describes a scalable Virtual Cluster Manager that provides such application agnostic cluster management capabilities to the system administrators maintaining virtual clusters over Xen powered virtual nodes.

Computational science simulations have been used to enable discovery in a broad spectrum of application areas, these simulations show irregular demanding characteristics of computing resources from time to time. The adoption of virtualized high performance cloud, rather than CPU-centric computing platform (such as supercomputers), is gaining interest of interests mainly due to its ease-of-use, multi-tenancy and flexibility. Basically, provisioning a virtual cluster, which consists of a lot of virtual machines, in a real-time has a critical impact on the successful deployment of the virtualized HPC clouds for computational science simulations. However, the cost of concurrently creating many virtual machines in constructing a virtual cluster can be as much as two orders of magnitude worse than expected. One of the main factors in this bottleneck is the time spent to create the virtual images for the virtual machines. In this paper, we propose a novel technique to minimize the creation time of virtual machine images and improve I/O performance of the provisioned virtual clusters. We also confirm that our proposed technique outperforms the conventional ones using various sets of experiments.

Hadoop MapReduce as a service from cloud is widely used by various research, and commercial communities. Hadoop MapReduce is typically offered as a service hosted on virtualized environment in Cloud Data-Center. Cluster of virtual machines for MapReduce is placed across racks in Cloud Data-Center to achieve fault tolerance. But, it negatively introduces dynamic/heterogeneous performance for virtual machines due to hardware heterogeneity and co-located virtual machine's interference, which cause varying latency for same task. Alongside, curbing number of intermediate records and placing reduce tasks on right virtual node are also important to minimize MapReduce job latency further. In this paper, we introduce Multi-Level Per Node Combiner to minimize the number of intermediate records and Dynamic Ranking based MapReduce Job Scheduler to place reduce tasks on right virtual machine to minimize MapReduce job latency by exploiting dynamic performance of virtual machines. To experiment and evaluate, we launched 29 virtual machines hosted in eight different physical machines to run wordcount job on PUMA dataset. Our proposed methodology improves overall job latency up to 33% for wordcount job.

To address the virtual machine cluster deployment issues in cloud computing environment, a novel MCSA (Min-cut segmentation algorithm) of virtual machine cluster is proposed with resource and communication bandwidth constraints. In this paper, the basic idea is based on the fully consideration on the CPU, memory, hard-disk and other resource constraints between virtual machine cluster and physical host, as well as the communication bandwidth constraints between the virtual machine. We quantified the virtual machine cluster constructed an undirected graph. In the undirected graph, the nodes represent the virtual machine, so the weight of a node represents the value of resources, and the edges represent the communication bandwidth, so the weight of the edge represents the value of communication bandwidth. Base on the above transformations, the resources and bandwidth constrained optimization problem is transformed into the graph segmentation problem. Next we segment the undirected graph by minimum cut algorithm, and computing the matching degree of physical machines. Last we obtained the approximate solution. To validate the effectiveness of the new algorithm, we carried out extensive experiments based on the CloudSim platform.

The rapid development of cloud computing and high requirements of operators requires strong support from the underlying Data Center Networks. Therefore, the effectiveness of using resources in the data center networks becomes a point of concern for operators and material for research. In this paper, we discuss the online virtual-cluster provision problem for multiple tenants with an aim to decide when and where the virtual cluster should be placed in a data center network. Our objective is maximizing the total revenue for the data center networks under the constraints. In order to solve this problem, this paper divides it into two parts: online multi-tenancy scheduling and virtual cluster placement. The first part aims to determine the scheduling orders for the multiple tenants, and the second part aims to determine the locations of virtual machines. We first approach the problem by using the variational inequality model and discuss the existence of the optimal solution. After that, we prove that provisioning virtual clusters for a multi-tenant data center network that maximizes revenue is NP-hard. Due to the complexity of this problem, an efficient heuristic algorithm OMS (Online Multi-tenancy Scheduling) is proposed to solve the online multi-tenancy scheduling problem. We further explore the virtual cluster placement problem based on the OMS and propose a novel algorithm during the virtual machine placement. We evaluate our algorithms through a series of simulations, and the simulations results demonstrate that OMS can significantly increase the efficiency and total revenue for the data centers.

The modeling and dependability analysis of virtualization system is always challenging due to the particular characteristic and mechanism of virtualization, in this paper we attempt to modeling virtual clusters using a hierarchical method combining a upper-layer Reliability Block Diagram (RBD) model and a lower-layer Markov model. Taking into account both the failure of physical servers and the virtual machines, the Markov model for one cluster is established while the RBD model reveals an overall structure of virtual clusters as well. The dependability of which three attributes are concentrated including availability, reliability and maintainability in this paper, is then theoretically analyzed by solving the proposed hybrid model, some useful advices are suggested to design an effective virtualization system with required dependability in according to the numerical results.

Resource management strategy in adaptive cloud provisions the needed resources dynamically to the end-users. To improve the runtime performance of adaptive cloud for service-based applications, two aspects of technical issues are required to be addressed. The first one is the balancing of a large amount of data on existing resources and the second is resource provisioning which can adjust the number of resources optimally to adapt the time-varying workload. As the growth of data is increasing tremendously, efficient resource management is the need in cloud computing. We build a cloud framework to process data in automation with adaptive resource and workload management strategy. Numbers of approaches are reviewed and applied for workload prediction. We developed the model Auto-Regressive Integrated Moving Average-workload Prediction for Efficient Resource Provisioning (ARIMA-PERP) and evaluated the results that can satisfy the on-demand need of end-users for efficient resource utilization. To serve the maximum number of user requests, performance metrics of the proposed approach are evaluated. It is observed that our evaluated results achieved an accurate prediction by 91.11%, which meets the efficient resource utilization for the demanded workload. As compared with the exiting approach, we achieved better performance by 0.11% for accurate prediction. The proposed architecture is intended to provide the resources dynamically and efficiently satisfying the demands of the user. To achieve this objective of efficient resource provisioning, algorithms are developed for workload prediction which helps in deciding optimum resource provisioning. Our system uses proactive approach resource management and deployment of the adaptive cloud system. In traditional systems, resources are managed based on demand, availability and the strategy of scheduling which results in delayed response time at large. We configured the ARIMA model to predict the future workload for provisioning the resources dynamically and remove the problem of over-provisioning and under-provisioning in a cloud environment.

In modern computing environments such as cloud, edge, and fog computing, as well as IoT networks and real-time systems, meeting workload deadlines is critical to ensure reliability, quality of service, and user satisfaction. The traditional scheduling algorithms often fail to adequately address the constraints associated with the workloads, particularly deadlines, the dynamic nature of workloads and resources, and the inherent resource limitations. Deadlines are the most important constraints, especially for time-sensitive applications. Achieving deadline compliance requires optimal resource provisioning, scheduling, resource allocation, resource scaling, workload migration, etc. This paper proposes a novel deadline-aware cloudlet scheduling algorithm, the Two-Stage Deadline-Aware Cloudlet Scheduling Algorithm (TSDACS), designed to minimize deadline misses through efficient resource provisioning and scaling strategies. In stage one, the algorithm provisions virtual machines with suitable configurations and quantities based on the requirements of the cloudlets to ensure they can be processed within their deadlines. Cloudlets are then scheduled onto the virtual machines in a way that minimizes deadline violations. In stage two, if the initially provisioned virtual machines fail to meet the deadlines, horizontal scaling is applied, up to a limited threshold, to enhance the performance and the deadline compliance. Experimental results demonstrate that the proposed TSDACS algorithm outperforms existing approaches, such as CPDALB, DBS, and RDLBS2, in terms of deadline miss ratio, makespan, response time, and cost efficiency, while maintaining competitive VM utilization and effective load balancing.

<p>The purpose of this paper is to provision the on demand resources to the end users as per their need using prediction method in cloud computing environment. The provisioning of virtualized resources to cloud consumers according to their need is a crucial step in the deployment of applications on the cloud. However, the dynamical management of resources for variable workloads remains a challenging problem for cloud providers. This problem can be solved by using a prediction based adaptive resource provisioning mechanism, which can estimate the upcoming resource demands of applications. The present research introduces a prediction based resource provisioning model for the allocation of resources in advance. The proposed approach facilitates the release of unused resources in the pool with quality of service (QoS), which is defined based on prediction model to perform the allocation of resources in advance. In this work, the model is used to determine the future workload prediction for user requests on web servers, and its impact toward achieving efficient resource provisioning in terms of resource exploitation and QoS. The main contribution of this paper is to develop the prediction model for efficient and dynamic resource provisioning to meet the requirements of end users.</p>

This paper presents an architecture for service hosting on virtual clusters spanning multiple administrative domains that balances the requirements of application developers and resource provider system administrators. The presented architecture and implementation use virtual machines to simplify the deployment of externally-accessible persistent Web and Grid services while allowing resource provider system administrators to monitor hosted virtual machines and perform critical maintenance when necessary. This approach allows developers full control of their distributed resources, specifies a mechanism for resource provider monitoring and intervention, and reduces the barrier for hosting user-supplied virtual machines on shared resource provider cyberinfrastructure.