Concentration on large-scale systems, 1958 to 1963 (continued)

During this time (1959-62), Pearcey and Hill surveyed the computing requirements of CSIRO and some technical Commonwealth agencies. Pearcey concluded it would be least expensive to serve those needs by a multi-level network of computers sited at most appropriate places, so that the more frequent computations requiring less computing power would be carried out locally while the larger jobs would be transmitted to, and performed by, one or other more powerful machines located elsewhere. The user would input his task locally and need not be aware of where it was actually executed. At that time transmission of programs and data by telecommunications at the level required was not feasible (Smart, K. J., 1960) and inter-level communication had, at least initially, to depend upon the loading on to tape of those jobs to be transferred by airfreight for execution elsewhere and similar return of outputs. A job turnaround time of twenty-four hours was expected and achieved.

A proposal for establishment by CSIRO of a Section of Computing Research, centred at Canberra, to operate a computing network service and engage in software research was accepted and funded by a Commonwealth Government grant of £3,000,000. The network on the lines described above was in operation by the end of 1963 using Control Data 3600 and 3200 systems. About the same time a similar scheme, using similar equipments but less dependent upon networking, was established by the Commonwealth Bureau of Census and Statistics. It was intended that the CSIRO network would, as soon as possible, be converted to on-line use via digital telecommunications. Conversion began in 1966 (Div. Comp. Res., 1967) and the network has since grown into the present CSIRONET, with 150 nodes throughout Australia.

One of the first of the Section's research projects was the development by J. B. Austin, T. S. Holden, P. Frost, H. H. Hudson and T. Pearcey (Frost, P. J. and Pearcey, T., 1966) and with valuable assistance from Control Data's staff of an on-line, multi-user system for the 3600, called DAD, that allowed time-shared access by many users using keyboard VDU terminals (which were then new). This was the first (1965) large scale, on-line, multi-user system in Australia other than the small Cirrus.

Australian Academy of Technological Sciences and Engineering