The VIEW Project

Project Information

Discipline
Computer Science (401) 
Orientation
Research 
Abstract

Cloud computing has rapidly become a new platform featuring scalability and elasticity for executing business and scientific applications. This project aims to investigating Nimbus and Eucalyptus as Cloud platforms for elastic workflow scheduling and resource provisioning.

Intellectual Merit

The intellectual merits of this research lie in the following transformational contributions to the fields of scientific workflows and Cloud computing: 1) The development of a parallel data model to support the modeling, storage, querying, and processing of collection-oriented scientific data in scientific workflows, in which data parallelism is natively supported. 2) The development a MapReduce-enabled workflow model to support MapReduce-style scientific workflows in which each map or reduce job can be a scientific workflow itself.

Broader Impacts

The success of this project will provide a general-purpose but domain customizable, Cloud-oriented scientific workflow tool for accelerating scientific discovery. The PI is organizing an annual IEEE International workshop on scientific workflows to distribute the expected tool to various domains. The research results will also be broadly disseminated through publications in international journals and conferences.

Project Contact

Project Lead
Shiyong Lu (shiyong) 
Project Manager
Shiyong Lu (shiyong) 
Project Members
Fahima Bhuyan  

Resource Requirements

Hardware Systems
  • hotel (IBM iDataPlex at U Chicago)
  • india (IBM iDataPlex at IU)
  • sierra (IBM iDataPlex at SDSC)
  • xray (Cray XM5 at IU)
  • I don't care (what I really need is a software environment and I don't care where it runs)
 
Use of FutureGrid

We will use FutureGrid to provision Cloud resources for experimenting and evaluating various Cloud-based scientific workflows scheduling algorithms that are currently being developed in our lab.

Scale of Use

We will need around 10 VMs initially for experiments and testing with 10 hours per week. After 3 months, we expect the workload will be 20 hours per week. At the end of the year, we might need do a few tests that need 100VMs, but will run for only a short time (10-20 hours).

Project Timeline

Submitted
05/26/2011 - 14:02 
»