Dispersed Computing

Abstract

The Dispersed Computing program is developing techniques to distribute computing tasks across network computing elements to enable more efficient utilization of enterprise and Internet-based storage, processing, and networking resources. At present, enterprises and Internet-based information technology service providers are increasingly adopting the cloud model, with data storage and computer processing concentrated in large data centers, which brings economies of scale and cost savings to storage and processing, but creates problems for the network and for latency-sensitive applications due to the need to backhaul data to (often distant) data centers for processing. The Dispersed Computing program will develop a dispersed computing architecture that results in more efficient utilization of storage, processing, and networking resources. A key enabler is the recent introduction by vendors of network elements that can be dual-purposed as computational elements. These dual-purposed network-compute elements make it possible to eliminate bottlenecks/chokepoints and to mitigate impossible backhaul requirements by opportunistically moving code to data given network conditions and available network-compute elements. With Dispersed Computing technology, the network becomes the cloud and computation is performed where it is most efficient to do so.

Document Details

Document Type
Accomplishment
Publication Date
Oct 01, 2020
Source ID
aa4aeaff767017caaa6923fac406a21e

Tags

Fields of Study

  • Computer science

Readers

  • Computer Networking
  • Economics
  • Parallel and Distributed Computing.

Related Documents