Ein Metascheduler zur Skalierung vom Grid zum Cloud

From Lsdf
Revision as of 15:40, 1 October 2013 by Tao (talk | contribs) (Created page with "Zurück zur Themenliste = Title: Developing a Meta-scheduler for Scaling Grid to Cloud = = Overview = Cloud computing [1] [2] is a novel compu…")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Zurück zur Themenliste

Title: Developing a Meta-scheduler for Scaling Grid to Cloud


Cloud computing [1] [2] is a novel computing paradigm. It provisions on-demand computing resources in an easy-to-use way, attracting more and more users to run their applications on the Cloud. The large resource pool of computing Clouds is definitely an additional target for Grid users to run their scientific jobs. However, Cloud Computing adopts a business model “pay-as-you-use”, i.e. users must pay for their CPU hours on the Cloud. In contrast, Grid is normally free for individual users. In this case, a user has to make choices – either running the jobs on the Cloud with payment or keeping the jobs in the job queue to wait for execution. An automatic mechanism that can automatically make the decision will be useful to the users for this situation.

Tasks of the master thesis

The master thesis aims at developing a meta-scheduler that automatically schedules the Grid jobs to the Cloud in case that the Grid is overloaded, indicating a long wait time in the job queue. The meta-scheduler has to collect information about the resource status of the underlying Grid infrastructure and analyze the job queue to estimate the waiting time of the Grid jobs, to which users have a willing to pay for running on the Cloud. Since it is not easy o get such information in a real grid infrastructure, this development work will be performed on GridSim [3], a simulator for Grid Computing. Similarly, the execution on the Cloud can be modeled with CloudSim [4]. The whole work contains the following tasks:

  • Research survey.
  • Design and Implementation of the scheduling mechanism
  • Performance evaluation.
  • Write-up thesis and scientific paper.


The work requires background knowledge about parallel and distributed computing, such as Grid and Cloud computing system, and programming skill in Java.


  • [1] M. Armbrust, A. Fox, R. Griffith, et al. Above the Clouds: A Berkeley View of Cloud Computing.
     Technical report, University of California at Berkeley, February 2009. 
  • [2] Amazon EC2: http://aws.amazon.com/ec2/
  • [3] GridSim
  • [4] CloudSim: A Framework for Modeling and Simulation of Cloud Computing Infrastructures and
     Services. http://www.cloudbus.org/cloudsim/