Performance Of Cloud Services Research Articles

Dynamic Request Splitting for Interactive Cloud Applications

Deploying interactive applications in the cloud is a challenge due to the high variability in performance of cloud services. In this paper, we present Dealer — a system that helps geo-distributed, interactive and multi-tier applications meet their stringent requirements on response time despite such variability. Our approach is motivated by the fact that, at any time, only a small number of application components of large multi-tier applications experience poor performance. Dealer continually monitors the performance of individual components and communication latencies between them to build a global view of the application. In serving any given request, Dealer seeks to minimize user response times by picking the best combination of replicas (potentially located across different data centers). While Dealer requires modifications to application code, we show the changes required are modest. Our evaluations on two multi-tier applications using real cloud deployments indicate the 90%ile of response times could be reduced by more than a factor of 6 under natural cloud dynamics. Our results indicate the cost of inter-data-center traffic with Dealer is minor, and that Dealer can in fact be used to reduce the overall operational costs of applications by up to 15% by leveraging the difference in billing plans of cloud instances.

Applying Design of Experiments (DOE) to Performance Evaluation of Commercial Cloud Services

Appropriate performance evaluations of commercial Cloud services are crucial and beneficial for both customers and providers to understand the service runtime, while suitable experimental design and analysis would be vital for practical evaluation implementations. However, there seems to be a lack of effective methods for Cloud services performance evaluation. For example, in most of the existing evaluation studies, experimental factors (also called parameters or variables) were considered randomly and intuitively, experimental sample sizes were determined on the fly, and few experimental results were comprehensively analyzed. To address these issues, the authors suggest applying Design of Experiments (DOE) to Cloud services evaluation. To facilitate applying DOE techniques, this paper introduces an experimental factor framework and a set of DOE application scenarios. As such, new evaluators can explore and conveniently adapt our work to their own experiments for performance evaluation of commercial Cloud services.

Log-based Cloud Service Performance Maintenance Model

Log-based Cloud Service Performance Maintenance Model

Networking for the Cloud: Challenges and Trends

Cloud services have changed the way computing power is delivered to customers, by offering computing and storage capacity in remote data centers on demand over the Internet. The success of the cloud model, however, has not come without challenges. Cloud providers have repeatedly been related to reports of major failures, including outages and performance degradation. The internal network of cloud data centers has frequently been identified as a root-cause of these problems, showing that network provisioning and monitoring is still a major challenge for the deployment of cloud services. This paper argues that today’s technologies for measuring and monitoring Internet traffic could be applied in the context of the internal network of cloud data centers as well. To support that, we first show the suitability of flow-based traffic measurements for monitoring cloud services. Then, we present a case on bandwidth capacity provisioning to exemplify how flow-based measurements can be used to guarantee the performance of cloud services. Finally, we discuss future directions we believe will guide the development of new cloud services. We advocate that next generation cloud services will not only rely on the Internet as a means to reach users, but also influence how the Internet itself is organized. We illustrate this trend by describing our ongoing research on mobile clouds.

Avoiding the Disk Bottleneck in Deduplicated VM Transfer

This paper presents an optimization mechanism to increase the performance of cloud services that transfer groups of deduplicated virtual machine (VM) images. This is necessary as the naive data transfer approach for groups of deduplicated VM images is extremely inefficient as it generates highly random disk access pattern. The optimization mechanism presented significantly improves the VM image transfer performance through intelligent buffering, block-level transfer scheduling, and data indexing to increase disk access sequentiality.

Cloud mobile gaming

With the evolution of mobile devices and networks, and the growing trend of mobile Internet access, rich multi-player gaming using mobile devices, similar to PC-based Internet games, has tremendous potential and interest. However, the current client-server architecture for PC-based Internet games, where most of the storage and computational burden of the game lies with the client device, does not work with mobile devices, constraining mobile gaming to either downloadable, single player games, or very light non-interactive versions of the rich multi-player Internet games. In this article, we study a cloud server based mobile gaming approach, termed Cloud Mobile Gaming (CMG), where the burden of executing the gaming engine is put on cloud servers, and the mobile devices just communicate the users' gaming commands to the servers. We analyze the factors affecting the Quality of user Experience (QoE) using the CMG approach, including the game genres, video settings, the conditions of server and client, and the conditions of the wireless network. A Mobile Gaming User Experience (MGUE) model is developed and validated through controlled subjective testing. We also develop a prototype for in-service measuring MGUE, and then use it to characterize user experience achievable using the CMG approach in a mobile cellular network. The MGUE model developed and validated in this article will be helpful for researchers and service providers to develop and assess the performance of future cloud mobile gaming techniques and services.

Distributed wide-area traffic management for cloud services

The performance of interactive cloud services depends heavily on which data centers handle client requests, and which wide-area paths carry traffic. While making these decisions, cloud service providers also need to weigh operational considerations like electricity and bandwidth costs, and balancing server loads across replicas. We argue that selecting data centers and network routes independently, as is common in today's services, can lead to much lower performance or higher costs than a coordinated decision. However, fine-grained joint control of two large distributed systems---e.g., DNS-based replica-mapping and data center multi-homed routing---can be administratively challenging. In this paper, we introduce the design of a system that jointly optimizes replica-mapping and multi-homed routing, while retaining the functional separation that exists between them today. We show how to construct a provably optimal distributed solution implemented through local computations and message exchanges between the mapping and routing systems.

The Performance Prediction of Cloud Service via JOGM(1,1) Model

With the rapid development of cloud computing, more and more cloud services are provided to users. Faced with multiple cloud services, how to scientifically predict and assess the performance of cloud service is an imperative task. To cope with the challenge, an prediction scheme via JOGM(1,1) model is proposed based on grey system theory, in which the performance of cloud service is quantified as the response time. Thus cloud users and the third-party institutes of cloud service performance evaluation can predict and assess the performance of cloud service as accurate as possible. In return, this will contribute to the cloud service provider selection for better performance on behalf of cloud consumers. The simulation results show that the proposed scheme has higher prediction precision compared with classic GM(1,1) model and weighted moving average model, which verifies the effectiveness of the proposed prediction scheme and the feasibility of forecasting the performance of cloud service employing grey system theory.

클라우드 컴퓨팅 서비스의 가용성 최적화를 위한 모델링 및 시뮬레이션

Cloud computing emerges as a new paradigm for deploying, managing and offering IT resources as a service anytime, anywhere on any devices. Cloud computing data center stores many IT resources through resource integration. So cloud computing system has to be designed by technology and policy to make effective use of IT resources. In other words, cloud vendor has to provide high quality services to all user and mitigate the dissipation of IT resources. However, vendors need to predict the performance of cloud services and the use of IT resources before releasing cloud service. For solving the problem, this research presents cloud service modeling on network environment and evaluation index for availability optimization of cloud service. We also study how to optimize an amount of requested cloud service and performance of datacenter using CloudSim toolkit.

Performance evaluation of cloud service considering fault recovery

Cloud computing is a recent trend in IT, which has attracted lots of attention. In cloud computing, service reliability and service performance are two important issues. To improve cloud service re...