A peer-to-peer solution for scalable virtual application deployment and execution on demand

This technology is a method for rapidly distributing and providing access to stored content from remote storage.

Unmet Need: Efficient virtual machine distribution for cloud applications

As more and more computing applications transition to the cloud, there is a need for a rapid and scalable approach to virtualization. A common bottleneck in virtualization is task execution. Existing virtual appliance deployment software requires full download of the virtual image before execution of the task, which leads to delays.

The Technology: Intelligent data retrieval for quick and scalable virtual machine distribution

This technology provides a rapid and scalable approach to virtualization by dividing the required virtual machine software image into smaller downloads. It utilizes a peer-to-peer network to retrieve the needed modules on demand, fetching first the modules needed for execution. The technology further minimizes delays by prefetching data modules based on a predictive algorithm of known task execution patterns and dynamically reacting to real-time demand from the host.

The technology has been implemented and tested using a variety Linux and Windows desktop virtual appliances. It is able to execute many workloads using only 5-15% of the full virtual image download through the use of module segmentation and efficiently conserves bandwidth through the use of its predictive algorithm.

Applications:

  • Enterprise-wide software deployment
  • Datacenter distribution services
  • Applications in the cloud
  • Peer-to-peer network optimization
  • Linux- and Windows-based virtual machine monitors

Advantages:

  • Scalable distribution of identical virtual machines
  • Rapid on-demand data retrieval
  • Predictive and dynamic data retrieval for bandwidth optimization in peer-to-peer setting

Lead Inventor:

Vishal Misra, Ph.D.

Patent Information:

Patent Status

Related Publications:

Tech Ventures Reference: