Over the past seven years, the Jefferson Lab’s control system has grown to include more than two hundred distributed computers running over a complex segmented network, controlling a number of semi-independent operational plants. Several of the plants, in cluding that used for running beam for physics users, operate around the clock with only brief, scheduled interruptions for machine repairs. Because of this, high control system availability is critical. Dividing computing resources into distinct sections , called fiefdoms, improves availability of the control system for each plant while facilitating periodic maintenance. In order to maximize uptime, each fiefdom operates as a completely independent control system consisting of a file server machine with a complete set of control system software and files, a local network, operator consoles and computers to execute high and low level control programs. The fiefdoms are isolated using network hardware, while still allowing limited communication between them. By segmenting the control system in this manner, the effect of a computer failure is minimized and machines can be taken down for periodic maintenance and upgrades without disabling other controls capabilities for the site.

This work was supported by U.S. DOE under contract DE-AC05-84ER40150.

Authors: K.S. White, M. Bickley

Abstract submitted to PAC 2001, Chicago, IL, June 18-24, 2001

Updated March 28, 2001

maintained by the Publications Manager