, IEEE Computer Society
Pages: pp. 1058-1059
Cloud Computing is defined as a pool of virtualized computer resources. Based on this virtualization, the Cloud Computing paradigm allows workloads to be deployed and scaled-out quickly through the rapid provisioning of virtual machines or physical machines. A Cloud Computing platform supports redundant, self-recovering, highly scalable programming models that allow workloads to recover from many inevitable hardware/software failures and monitoring resource use in real time for providing physical and virtual servers, on which the applications can run. A Cloud Computing platform is more than a collection of computer resources, because it provides a mechanism to manage those resources. In a Cloud Computing platform, software is migrating from the desktop into the “clouds” of the Internet, promising users anytime, anywhere access to their programs and data.
In the paper “On the Optimal Allocation of Virtual Resources in Cloud Computing Networks,” by C. Papagianni, A. Leivadeas, S. Papavassiliou, V. Maglaris, C. Cervelló-Pastor, and Á. Monje, formulate the Virtual Network Embedding (VNE) problem in networked cloud environment.
Following cloud service paradigm, the paper aims to: 1) extend the pool of shared resources to a layer 2/3 network topology, including heterogeneous network infrastructure, possibly across multiple domains; 2) provide a generic formulation for the resource mapping problem at hand capable of taking into consideration Quality of Service (QoS) requirements; 3) support QoS provisioning of cloud Infrastructure as a Service (IaaS); 4) design and implementation of an experimentation simulation environment that allows a flexible and structured evaluation of the performance and efficiency of the proposed approach; and finally 5) provide a proof of concept, of the operational efficiency of the proposed approach, via a prototype implementation of the framework on an FI experimentation platform - FEDERICA.
In the paper “Workload-Based Software Rejuvenation in Cloud Systems,” by D. Bruneo, S. Distefano, F. Longo, A. Puliafito, and M. Scarpa, the main goal of time-based rejuvenation models is to find an optimal rejuvenation timer that allows to minimize some objective functions. Usually, the timer is set at a system start-up and it does not change with respect to the system dynamics (e.g., system workload variations). The authors refer to such kind of approach as fixed timer policy. Another contribution of the present work is the specification of a time-based policy adapting the rejuvenation timer to the Virtual Machine Monitor (VMM) conditions, taking into account its workload and ageing (variable timer policy). The effectiveness of the proposed modeling technique is demonstrated through a numerical example, based on a case study, taken from the literature. It shows how the proposed variable timer policy outperforms the fixed one in terms of improved system availability also varying the way failure rates are affected by the workload. It can be noted that the authors present an analytic technique that allows to represent any generic failure and repair distributions, adequately modeling changes in the workload through the conservation of a reliability principle.
The paper “Integrated Approach to Data Center Power Management,” by L. Ganesh, H. Weatherspoon, T. Marian, and K. Birman, focuses on a key aspect of data center operational efficiency-energy management.
This paper takes an integrated approach to data center energy management to simultaneously address idle resource energy consumption, and support-infrastructure energy consumption. The authors argue for a power management approach that powers down racks or even entire containerized data centers, when idle, thus powering down not only servers, but also their associated power distribution, backup, networking, and cooling equipment. The evaluation shows that shifting to this model combines the energy savings of the power-proportional as well as the green data center approaches, while not impacting performance. They also show that this shift is practical today at very low deployment cost, and that current data center trends strongly enable it.
The authors believe that an increasingly likely vision of the future of online services is one where a few infrastructure providers compete to host the world's services and data. They show that for an SaaS provider, existing data replication and placement policies fit the proposed large Power Cycle Unit (PCU ) model. Further, the authors show that an SaaS provider could provide storage options up to 16.5 percent cheaper by adopting rack-based power management, and tuning the number of replicas kept live. Finally, they examine another point in the design space-container farms. The authors show that, in this scenario, using entire containers as the PCU is practical, and leads to no performance penalty over node-based power management.