ILL - Networking

The first steps were linking the PDP11/40 of D17 in 1976 with a serial link which could serve as a terminal or be used to transfer the data using task-task communications programs (Tillier, ILL80T15T). The protocol allowed use of either the instrument DEC VT11 graphic display or a GT40 terminal of the PDP10. More important it served for data transfer using a check-summed protocol (Le Sourne, Tillier). The first function was not used extensively, but the transfers using a check-summed protocol (Le Sourne, Tillier) were very reliable. The system was replicated on D11 when a PDP11/40 replaced the NICOLE system. A third instrument transferring data using this ILL protocol was IN6.

It was evident that expanding this task-to-task communications feature was quite expensive on the limited central computer resources. For this reason, when DECnet was introduced as part of the NICOLE replacement, a data concentrator was acquired (PDP11/55) which received data from the ex-NICOLE instruments and continued to write data tapes for transfer to the central computer. With additional computing power the KL1091S received serial line data from many of the neutron beam instruments; a polling program looked for data transfer requests from several instruments. The DEC computer had two additional DN20 front-end PDP11/34s to handle DECnet and other external links.

For DECNET phase I only coaxial or synchronous RS422 links were possible; these point to point links were too costly to envisage for general use, and initially were limited to the NICOLE instruments and the PDP11/55 concentrator. D11, using a VAX11/730, had a DMF32 interface which allowed it to be connected to a similar interface on the VAX8600, allowing full DECNET operations. By the time D17 was updated DECnet had been modified to use serial lines, shrinking the cost of point-to-point connections at 19200 baud. The reduced cost then, and lower cpu overheads of end node rather than routing network licences, was an additional gain.

Initially ethernet was limited to a few instruments, starting with the diffractometer D19, but this was mainly used to offload data treatment to the VAX11/750 associated with the D19 project though the limited cpu capacity of the PDP11/24 control computer could not take full advantage of the increased network bandwidth. The network was extended in-house progressively, and once the VAX8600 was installed thin-wire ethernet (10Base2) was cabled through much of the institute, and DECnet was used quite extensively for remote terminal access and file transfer (1987 on).

Earlier, indirectly, through the local IN2P3 laboratory, it had been possible to access CERN by DECnet. Wide-area networking started in the mid 80s with a bi-sync link to the CICG, University of Grenoble, when the ILL joined EARN, the European Bitnet network (store and forward of mail) with the node name FRILL.

When the network was extended through the ILL to EMBL the name changed to FRILL51. The technology of WAN advanced rapidly, and Transpac X-25 access took over the earlier transport layer of bi-sync for EARN, and the ILL joined the French RENATER university X-25 backbone managed by France Télécom (now Orange). This in turn offered connectivity to the JANET X-25 university system in the UK, and the ILL VMS systems had JANET software installed.