Xinkui Zhao

BURSE: A Bursty and Self-Similar Workload Generator for Cloud Computing

As two of the most important characteristics of workloads, burstiness and self-similarity are gaining more and more attention. Workload generation, which is a key technique for performance analysis and simulations, has also attracted an increasing interest in cloud community in recent years. Though a large number of methods for synthetically generating bursty or self-similar workloads have been proposed in the literature, none of them can deal with workload generation with both of the two characteristics. In this paper, a configurable and intelligible synthetic generator (BURSE) is proposed for bursty and self-similar workloads in cloud computing based on a superposition of two-state Markov Modulated Poisson Processes (MMPP2s). The proposed generator can produce workloads with both specified intension of burstiness and self-similarity. Detailed experimental evaluation demonstrates the accuracy, robustness and good applicability of BURSE.

System resource utilization analysis and prediction for cloud based applications under bursty workloads

Abstract Performance analysis and prediction need a solid understanding of the system workload. As a salient workload characteristic, burstiness has critical impact on resource provisioning and performance of cloud based applications. Thus performance analysis and prediction under bursty workloads are of crucial importance to cloud based applications. However, it is yet challenging for such analysis and prediction, since no accurate and effective bursty workload generator exists, as well as the fine-grained bursty workload analysis and prediction method. In this article, to deal with these challenges, a bursty workload generator has been proposed for Cloudstone (a cloud benchmark) based on 2-state Markovian Arrival Process (MAP2). Then based on this generator, a fine-grained performance analysis method, which can be used to predict the probability density function of CPU utilization, has been suggested for cloud based applications, to support better resource provisioning decision making and system performance optimization. Finally, extensive experiments are conducted in a Xen-based virtualized environment to evaluate the accuracy and effectiveness of the two methods. By comparing the actual value of Indices of Dispersion for Count with the target value deduced from MAP2 model, the experiments show the precision of our method is superior to existing works. By comparing the real and predicted system resource utilization under a variety of bursty workloads generated by the proposed generator, the experiments also demonstrate the effectiveness and accuracy of the proposed fine-grained system resource utilization prediction method.

CloudScout: A Non-Intrusive Approach to Service Dependency Discovery

Nowadays, numerous enterprises are migrating their applications into cloud computing environments. Typically, the applications are composed of several dependent service components that span many hosts and network devices. In light of this, exploring the dependency between service components can be beneficial for achieving fast network application response time. Moreover, it is significant to consolidate service components according to resource constraints, service dependency, and network structure. However, it is a tedious task to discover the dependency among service components without expert knowledge of the running application. In this paper, we propose CloudScout, a non-intrusive approach that is capable of automatically discovering dependent service components. CloudScout analyzes the correlation among service components based on the time-series information from system monitoring logs. We address two key challenges in CloudScout: service distance calculation and dependent service clustering. We conduct experiments on five applications with 290 service components that span 20 physical hosts across two data centers. The experimental results demonstrate that CloudScout can successfully discover the dependency among service components and facilitate reducing the network latency of network applications and distributed applications.

An Approach for Bursty and Self-similar Workload Generation

As two of the most important characteristics of Web systems’ workloads, burstiness and self-similarity are gaining more and more attentions. And synthetically generating bursty and self-similar workloads is a key technique for Web system performance analysis. In this paper, a configurable synthetic approach for bursty and self-similar workload generation has been proposed based on a superposition of 2-state Markovian arrival processes (MAP2). This method can generate workload with both specified intension of burstiness and self-similarity. The detailed evaluation show the accuracy and robustness of our method.

Workload Classification Model for Specializing Virtual Machine Operating System

There is growing demand on strategies to help cloud computing utilize its scale adaptiveness and cost effectiveness advantages. Previous operating systems(OS) are designed to suit all, leading to that virtual machines with different workloads use indiscriminate processing platform. However, there are conflicts between generality and performance, limited resource utilization and low processing efficiency of common OS penalize system performance. Therefore, we design four kinds of OS optimization strategies corresponding to four primary classes of workloads: CPU-Intensive, Memory-Intensive, I/O-Intensive and Network-Intensive. In this paper, we propose a Feedback-Based Workload Classification(FBWC) model which contains metrics collector, data preprocessor, Training Set Refresh Support Vector Machine(TSRSVM) classifier, decision maker and operating system tuner to classify workloads into appropriate class. TSRSVM combines support vectors of origin training set and correctly classified testing set together as new training set to get higher classification accuracy and efficiency. Comprehensive experiments compared with K Nearest Neighbors(KNN) and SVM demonstrate effectiveness of FBWC model and TSRSVM classification algorithm. Performance comparison between common virtual machine and the tuned one shows high degree performance improvement by OS specialization.

Distance-aware virtual cluster performance optimization: A hadoop case study

Cloud computing and big data are becoming two important developing trends in information technology area. However, data-intensive computing has some challenges to work well on virtual machines in cloud computing for virtualized resource competition and complex network communication. Network becomes one of the most notorious bottlenecks, which highlights strategies to lower communication and transmission cost in virtual cluster. In this paper, we present a novel cluster performance optimization strategy named vClusterOpt. vClusterOpt finds out centralized subgraphs of node graph and choose node with the shortest logical distance as kernel node of the subgraph to reduce inter-machine communication and transmission cost under virtual cluster. To calculate logical distance accurately, we define two kinds of logical distance: Logical Communication Distance(LCD) and Logical Transmission Distance(LTD). VM with the shortest LCD with others is used as the communication kernel node who has the most information communication stress, while VM with the shortest LTD is treated as transmission kernel node who has the most data transmission stress. We choose benchmarks running on Hadoop as the represent of data-intensive computing service to demonstrate effectiveness of our approach. Experiments show that an average of 20% performance improvement can get by our distance-aware virtual cluster optimization strategy.

SimMon: A Toolkit for Simulating Monitoring Mechanism in Cloud Computing Environments

Monitoring is significant to supervise the state of services and guide adaptive management of services in cloud computing environments. Working as auxiliary tools, monitoring systems are expected to incur the least extra cost on physical resources (CPU, memory, network, etc.). Since the scale and requirement of different data centers vary from each other, it is impossible to design a suit-to-all monitoring solution for all the data centers. However, for a certain data center, it is hard to determine whether a predesign monitoring mechanism is well suited before the mechanism is deployed in a real production environment. To address these issues, we propose SimMon, a toolkit for simulating monitoring mechanism in cloud computing environments. SimMon is used to simulate the process on collection, dissemination, storage and requisition of monitoring data. With the help of SimMon, system administrators are able to compare different monitoring mechanisms and select the best one before it is adopted by a monitoring system in a real-world data center.

HiSML: A High-Level Integrated Service Monitoring Language

In this paper, we propose HiSML, a high-level integrated service monitoring language. The language is designed to build monitoring solutions for cloud computing platforms. The primary benefits of HiSML over existing monitoring tools are: 1) it is used to build the monitoring solution from scratch, and the monitored objects are specialized for the target platform, 2) it integrally monitors services in all layers of cloud computing platforms: infrastructure layer, platform layer and software layer, 3) it allows programmers to describe the dependency between monitored services to guide analysis on the collected data, 4) it allows programmers to manually store and backup the monitored data, 5) it supports hybrid programming with other programming languages to assist the adaptive management of cloud computing platforms.

SimMon: a toolkit for simulation of monitoring mechanisms in cloud computing environment

Monitoring is a precondition for intelligent management in cloud computing environment, such as dynamic resource allocation. Typically, working as an auxiliary tool, a monitoring system is expected to incur the least additional resource usage, thus the strategies to improve the efficiency of monitoring mechanisms become significant. Yet the scale and monitoring requirement of different data centers vary, we cannot determine whether a monitoring mechanism would work well in a new data center before it serves the data center. To evaluate monitoring mechanisms, we propose SimMon, a toolkit for simulating monitoring mechanisms in cloud computing environments. SimMon is designed to simulate the topologies, actions, and strategies in data collection, dissemination, storage, and management processes. SimMon provides a controllable and repeatable way to evaluate monitoring mechanisms. In this paper, we describe the requirements analysis, design, implementation, and evaluation of SimMon. We simulate several different monitoring systems and compare their cost on time and resource to evaluate the efficiency of SimMon. We reproduce two usage scenarios from former literatures to demonstrate the effectiveness of SimMon on monitoring mechanisms simulation and evaluation. We build a real‐world working environment to validate the capability of SimMon on mimicking the characteristics of cloud monitoring systems. Copyright

vSpec: workload-adaptive operating system specialization for virtual machines in cloud computing

In general, operating systems (OSs) are designed to mediate access to device hardware by applications. They process different kinds of system calls using an indiscriminate kernel with the same configuration. Applications in cloud computing platforms are constructed from service components. Each of the service components is assigned separately to an individual virtual machine (VM), which leads to homogeneous system calls on each VM. In addition, the requirements for kernel function and configuration of system parameters from different VMs are different. Therefore, the suit-to-all design incurs an unnecessary performance overhead and restricts the OS’s processing capacity in cloud computing. In this paper, we propose an adaptive model for cloud computing to resolve the conflict between generality and performance. Our model adaptively specializes the OS of a VM according to the resource-consuming characteristics of workloads on the VM. We implement a prototype of the adaptive model, vSpec. There are five classes of VM: CPU-intensive, memory-intensive, I/O-intensive, networkintensive and compound, according to the resource-consuming characteristics of the workloads running on the VMs. vSpec specializes the OS of a VM according to the VM class. We perform comprehensive experiments to evaluate the effectiveness of vSpec on benchmarks and real-world applications.摘要创新点1、提出一种自动化负载特征选择方法, 根据互信息和皮尔森相关系数从监控指标中选择有效的特征;2、提出一种适用于负载分类的监督学习方法, 综合考虑已正确分类数据和将要分类数据之间的相似性, 提升分类准确率;3、提出基于负载分类的虚拟机操作系统定制化模型并实现参考原型系统, 对分类后的四种资源密集型应用采用对应的定制化方法优化应用所在的虚拟机的操作系统, 提升其运行性能。