Towards virtual networks for virtual machine grid computing

Ananth I. Sundararaj, Peter A. Dinda

Research output: Contribution to conferencePaperpeer-review

119 Scopus citations

Abstract

Virtual machines can greatly simplify wide-area distributed computing by lowering the level of abstraction to the benefit of both resource providers and users. Networking, however, can be a challenge because remote sites are loath to provide connectivity to any machine attached to the site network by outsiders. In response, we have developed a simple and efficient layer two virtual network tool that in effect connects the virtual machine to the home network of the user, making the connectivity problem identical to that faced by the user when connecting any new machine to his own network. We describe this tool and evaluate its performance in LAN and WAN environments. Next, we describe our plans to enhance it to become an adaptive virtual network that will dynamically modify its topology and routing rules in response to the offered traffic load of the virtual machines it supports and to the load of the underlying network. We formalize the adaptation problem induced by this scheme and take initial steps to solving it. The virtual network will also be able to use underlying resource reservation mechanisms on behalf of virtual machines. Both adaptation and reservation will work with existing, unmodified applications and operating systems.

Original languageEnglish (US)
StatePublished - 2004
Event3rd Virtual Machine Research and Technology Symposium, JAVA VM 2004 - San Jose, United States
Duration: May 6 2004May 7 2004

Conference

Conference3rd Virtual Machine Research and Technology Symposium, JAVA VM 2004
Country/TerritoryUnited States
CitySan Jose
Period5/6/045/7/04

ASJC Scopus subject areas

  • Computer Science Applications
  • Artificial Intelligence
  • Information Systems
  • Software

Fingerprint

Dive into the research topics of 'Towards virtual networks for virtual machine grid computing'. Together they form a unique fingerprint.

Cite this