Home
Members
Publications
Projects/Collab.
Software
Events
Open Positions
Contact
News
June 17, 2010 Thursday (Room George Besse) | |
---|---|
13:50 14:00 | Greeting |
14:00 15:00 |
Virtualization fundamentals This presentation will attempt to clarify and classify the virtualization landscape from a technical stand-point. Virtualization has a broader scope than data centers as will be shown by some use cases. We will introduce a classification to sort out the many usages of the "virtualization" word to distinguish JVM, VM, OS virtualization and many others. The most common ways to provide virtualization will be described and compared, both for basic resources (CPU, memory) but also for devices. Advanced virtualization capabilities will be also discussed. |
15:00 15:35 |
VMScript, a new DSL for VM manipulations Our goal is to propose VMScript, a domain specific language for administration of virtualized grid infrastructures. This language relies on set manipulation and is used to introspect physical and virtual grid architectures thanks to query expressions and notably to modify VM placement on machines. |
Coffee break | |
15:55 16:30 |
Virtualization for autonomic systems Our goal is to propose an autonomous decision center to be included in distributed system middleware. To accomplish this aim we use an autonomous management tool using a clever decision center able to manage virtual machines. Furthemore, for this we need to virtualize the monitoring and resources including processors, memory. We will present during the JTE this ongoing work. |
16:30 17:05 |
Virtual Machine based J2EE application management with TUNe One of the main services that must be assured in a hosting center is resource allocation. For economic reasons, resources should be allocated (and deallocated) dynamically to the hosted applications according to their runtime load. Autonomic computing systems provide support for such a dynamic resource allocation, as they allow applications to be dynamically reconfigured in order to involve more or less resources. We implemented an autonomic management system called TUNe that we used for dynamic resource allocation in a J2EE e-commerce application. We present the results of our experiment with different management policies and we pinpoint the advantages and limitations of this approach. |
17:05 18:00 |
Leveraging the Grid'5000 facility to study virtualization in distributed architectures The presentation will show how the Grid'5000 platform can help to study/address virtualization challenges. A kind of BOF will conclude this talk to discuss particular requirements (network issues, VM images management, Inter-sites Live migration...) |
June 18, 2010 Friday (Room Georges Besse) | |
---|---|
08:45 9:45 |
The Reservoir Project |
09:45 10:20 |
Investigating the Energy Consumption of Virtualised Servers This talk will discuss early results on using features from Xen hypervisor, such as virtual CPU throttling and pinning. It will demonstrate the impact on energy consumption of using these features and how they could be utilised when provisioning resources to a social calendar application.y |
Coffee break | |
10:45 11:20 |
HIPerNet, a network-resource virtualisation framework This presentation will discuss the motivations, concepts, and design of the HIPerNet framework. The key idea of HIPerNet is to combine network- and system-virtualization with controlled resource reservation for providing fully isolated environments. Experimental results obtained within the Grid'5000 testbed and a large-scale distributed application will be presented. Further information can be obtained at HIPCAL webpage. |
11:20 12:20 |
Large Scale Virtualization for Batch Services Virtualization has been studied extensively with a large uptake in recent years. Virtual machines provide application/user isolation, customization of the computing environment and flexibility via separation of concerns. The use of virtualization in grid computing has not been embraced as quickly as in the industry where virtualization is the key enabled of cloud computing. In this presentation we will present on-going work to virtualize the batch computing infrastrcuture of CERN. We will also show how it will be integrated with current grid computing middleware and infrastructure. In addition we will discuss a different architecture developed at Clemson University which uses a different hypervisor technology and a different networking setup than at CERN. The two will be compared. |
12:20 12:30 | Closing session |
Fill the registration form and send it by mail to seminaires-dre@mines-nantes.fr before June 11th, 2010 (13:00). Registrations received after this date will be accepted subject to availability.
'Full' package (including coffee breaks, the thursday dinner and the friday lunch): 60 Euros
'Sessions' package (including coffee breaks and the friday lunch): 20 Euros
Getting to the Ecole des Mines de Nantes: http://www.mines-nantes.fr/en/Pratique/Venir-a-l-Ecole