Expo/Science & Industry/Whispers From the Cosmos

| Back | Up | Map | Glossary | Information |

High Speed Networks

High speed data communication links are essential to future progress in radioastronomy. Enhanced connectivity will help improve the productivity of research by vastly increasing data transfer rates between instruments, computers and users.

Already, current network connections, such as the link between the Hat Creek array and the University of California at Berkeley, are proving far too slow for the large volumes of data they must carry. This bottleneck can only worsen when, as a result of its projected expansion, the BIMA array generates yet more data.

Lew Snyder, on-camera
Movie/Sound Byte
QuickTime Movie (1.5 MB); Sound File (864K); Text

One goal of the BIMA consortium is to deploy high speed networks that will serve astronomers throughout the United States, connecting them not only to the array, but also to supercomputers, a data archive and each other. The network will use advanced communications testbeds connecting the BIMA array and high performance computers in Illinois at NCSA . As part of the BLANCA networking testbed, high speed links were tested between NCSA and the University of Wisconsin for prototyping collaborative, realtime imaging and analysis between remote sites. Making its debut at Supercomputing '95, a new networking testbed, the Information Wide Area Year network (I-WAY), will be deployed in order to experiment further with computationally-intensive data reduction and imaging techniques.

From Network to a Metacomputer

Realtime astronomy demands that data be transmitted as quickly as it is processed and vice versa. If this match in speeds were realized, and if the interfaces and software for data handling and image generation were also standardized across all computer p latforms, then the network and all its components would appear to the local user as a single system, a metacomputer .

At all times, computations would seem as if they were being performed locally, on the office workstation, even though the most intensive calculations might be running on a remote supercomputer to which it is linked via a high speed network. For example, in future realtime steering of the telescope, researchers would send computationally heavy tasks, such as initial data reduction, to a distant supercomputer. Less demanding tasks, like rendering preliminary images for steering, would be handled locally, perhaps collaboratively. Glimpsing identical images on workstations connected by the same high speed network, scientists at many locations could collaborate, both when observing and later on, in analyzing the image output from the telescope.

Return to to Grand Challenge in Astronomy
Up to Computers: The Image Forming Elements

Exhibit Map
Information Center

Copyright © 1995: Board of Trustees, University of Illinois

NCSA. Last modified 11/13/95