Article Preview
TopCloud Architecture
In virtualized environments, the customers who purchase virtual machines (VMs) from a third-party cloud would expect that their VMs run in an isolated manner. However, the performance of a VM can be negatively affected by co-resident VMs. The paper by Ziye Yang, Haifeng Fang, Yingjun Wu, Chunqi Li and H.howie Huang, entitled “Measuring the Characteristics of Hypervisor I/O Scheduling in the Cloud for Virtual Machine Performance Interference”, presents a distributed I/O performance measurement system, which can help identify the characteristics of disk I/O scheduler in a hypervisor and conduct I/O based performance attacks. The authors conduct a number of experiments on both Xen and VMware platforms. The authors deploy their system on Amazon EC2 and successfully slow down the performance of co-resident VMs.
Along with the development of cloud computing, offloading has become an increasingly attractive way to extend the battery life and reduce execution time on mobile devices. The paper by Huaming Wu, Qiushi Wang and Katinka Wolter, entitled “Optimal Cloud-path Selection in Mobile Cloud Offloading Systems Based on QoS Criteria”, explores the methods of optimal cloud-path selection for offloading in mobile cloud computing systems when taking the network bandwidth between the mobile device and cloud service and the availability of cloud service into considering. Several alternative cloud services are considered and evaluated in terms of many different criteria such as performance, bandwidth, security, financial and availability in cloud-path selection problem. The proposed architecture is proved to be an effective and synthesized way through numerical analysis.
Resource Management
Data analytics in the Cloud opens new possibilities for executing complex applications with multiple processing phases. Such applications can benefit from using MapReduce model, only requiring the end-user to define the application algorithms for input data processing and the map and reduce functions. However, this poses a need to install/configure specific frameworks such as Apache Hadoop or Elastic MapReduce in Amazon Cloud. In order to provide more flexibility in defining and adjusting the application configurations, the paper by Carlos Goncalves, Luis Assuncao and Jose C. Cunha, entitled “Flexible MapReduce Workflows for Cloud Data Analytics”, describes an approach for supporting MapReduce stages as sub-workflows, termed as AWARD, which stands for Autonomic Workflow Activities Reconfigurable and Dynamic. The AWARD illustrates the feasibility of using a unified workflow framework to express the composition of different phases of a complex application. The authors have shown that one can execute a particular application phase as a sub-workflow designed as a MapReduce computation. They also showed that the proposed framework is flexible to support the execution of MapReduce workflows using similar APIs as used in Hadoop.
The Cloud computing paradigm is adopted for its several advantages like reduction of cost incurred when using a set of resources. However, despite the many proven benefits of using a Cloud infrastructure to run business processes, the lack of guidance for choosing between multiple offerings is still an open issue. The paper by Kahina Bessai, Samir Youcef, Ammar Oulamara and Claude Godart, entitled “Scheduling strategies for business process applications in Cloud environments”, proposes a set of scheduling strategies for scheduling business processes on distributed Cloud resources while taking into account its elastic computing characteristic that allows users to allocate and release compute resources on-demand. Experiment results demonstrate that the proposed approaches present good performance.