Investigation on Network Performance in Virtual Machine Based Cloud Environments

Project Information

Discipline
Computer Science (401) 
Orientation
Research 
Abstract

This proposed research will investigate the network performance of various cloud applications (e.g., live video streaming, video on demand, big data processing, etc.) in virtual machine based cloud environments. The investigation will first be done by measurement with real applications running on cloud platforms (e.g., FutureGrid). The measured data will then be analyzed to identify current limitations on the network performance and cloud platforms. Algorithms and solutions will then be designed to address the identified issues and improve the overall performance. The designed algorithms and solutions will be further evaluated by experiments on cloud platforms such as FutureGrid.

Intellectual Merit

Conduct measurements to investigate and analyze the network performance of various cloud applications on current cloud platforms. Design application-customized solutions to address identified issues and improve the overall performance.

Broader Impacts

The research results will be submitted for research publications, which allow the communities to better understand the network performance and issues on current cloud platforms, as well as provide guidelines for application-customized optimizations on network and overall performance for current and future cloud applications.

Project Contact

Project Lead
Feng Wang (fengwang) 
Project Manager
Feng Wang (fengwang) 
Project Members
Sampath Gowrishetty  

Resource Requirements

Hardware Systems
  • alamo (Dell optiplex at TACC)
  • foxtrot (IBM iDataPlex at UF)
  • hotel (IBM iDataPlex at U Chicago)
  • india (IBM iDataPlex at IU)
  • sierra (IBM iDataPlex at SDSC)
  • xray (Cray XM5 at IU)
  • bravo (large memory machine at IU)
  • delta (GPU Cloud)
  • Network Impairment Device
 
Use of FutureGrid

We will use FutureGrid for our measurements and analysis on the network performance of various cloud applications. We will also use FutureGrid as an experiment testbed for our designed algorithms and solutions.

Scale of Use

Most of time only a few VMs for measurements, analysis and performance evaluation. Sometimes may need more VMs to examine scalability issues, which will only take several hours each time.

Project Timeline

Submitted
12/18/2013 - 15:14 
»