Extending local resource using Clouds
Project Information
- Discipline
- Computer Science (401)
- Orientation
- Education
Our work is to establish an on-demand and dynamic extension of an existing cluster resource using clouds.In our model we build an additional layer which is separate from the applications and capable of measuring their workload and acquiring or releasing cloud according to the demand. A resource manager is built on the Nimbus toolkit to dynamically and securely extend existing physical clusters into the cloud.This resource manager interfaces directly with torque and parses its output to evaluate the status of the cluster nodes.The decision to add or release nodes from the cloud is made according to the policies and measuring the workloads handled by the cluster resources.
Intellectual MeritBuying resource with peak time usage in mind is too expensive. So hiring a cloud, but only using it whenever there is additional demand will pave for a new development in the industry. This will dramatically reduce cost in the case of Universities where, the traffic is too much during the publishing of results or during on-line examinations
Broader ImpactsThis project helps in getting access to resources at a remote place to a broader audience. Educational Institutions can make use of remote resources in computations, experiments and so is the scientific community. resource scarcity can be reduced with this effort
Project Contact
- Project Lead
- John Thottam (johnjose)
- Project Manager
- John Thottam (johnjose)
Resource Requirements
- Hardware System
-
- hotel (IBM iDataPlex at U Chicago)
We would like to deploy VMs upon demand which is to configured to join the physical cluster head node during the experiments.We need access to Nimbus based clouds which are available
Scale of UseWe need to launch a few VMs for the experiment which wont last for more than an hour during a single launch.
Project Timeline
- Submitted
- 02/20/2011 - 08:40