QANTAS

I started working for Honeywell in Brisbane in 1969 in their Computer Control Division (CCD). 

There was a few computer systems in Brisbane at the time, but by August 1969 we were in Sydney to work on the installation of a message switching system for QANTAS Airways which provided a front end to their Passenger Reservation Systems.

QANTAS acquired a number IBM S/360s in 1966 to continue development of its International Passenger Airways Reservation System (IPARS). My recollection is that all external messages for bookings and related information were by Telex. The Telex network was a switched-line world-wide network of Telex machines operating at 110 bps (at this time). The code was 5 bit and messages as received or for transmission were punched out on a 5 channel paper tape. The system employed a fairly simple address identifier and relied on a WRU (who are you) response when setting up a transmission.

QANTAS was the message switching centre for the South Pacific region for airline reservations and other related bookings. I do not have data for how many messages this system handled. There were a large number of Telex machines (my memory is foggy here but my estimate would be 30 to 50 machines)  dedicated to either sending or receiving messages. As messages were received operators would physically read the 5 channel address code on the message, tear off the message tape and splice it onto the end of the tape currently being sent to that destination. The tape for transmission was held loosely in a large rubbish bin at each machine. The operators were very skilled and often ran between machines performing their switching task. I don't recall what they did with messages bound for Qantas locally in Sydney. 

The Honeywell project aim was to automate the very labour intensive manual message handling process and to expand transmission to other and newer technologies. I do not have specific written details or diagrams of the hardware configuration or software developed but the Honeywell System Supervisory Panel outlines the project fairly well.

In addition to the Honeywell H632 (32 bit) machines, there were a number of H316 (16 bit) computers deployed locally and remotely. Both of these machine types had direct memory access which allowed peripherals to read/write information directly to/from memory simultaneously and independently of the CPU. This memory access method, together with a priority interrupt system reduced the CPU load considerably and rightly entitled Honeywell's claim that these machines were for real-time applications.

The H632 direct memory access was called the MAD (Memory Access Director). This method allowed simultaneous access of up to 4 banks of memory.  The front panel here shows the extended MAD which had a total of 16 channels. Each H632 had a CPU (Central Processor) and an IOP (Input/Output Processor) which could use a shared MAD memory channel or each have its own channel. 

The connection to the IBM S/360 was via a Honeywell adaptor which connected to the IBM Multi-Channel Bus that was used for slower speed peripherals such as card readers, printers, etc (the S/360 also has a high speed synchronous IO bus). These interfaces were electrically isolated from the IBM system and on the H632 side connected to the MAD to enable direct access to memory. I have the recollection that each of these adaptors was based on an H316.

Finally, the real world front-end to the H632s was provided by an H316 machine which connected to dedicated and switched (Telex) serial lines. This interface most likely used a 64 line Multi Line Controller designed by BBN (Bolt Beranek and Newman) for use on the H316 IMP (Interface Message Processor) terminal machines used in ARPANET. 

The 16 Channel MAD was the smart enabler for this project and provided a high speed switching capability between up to 4 IBM S/360s and the outside world. 

Project Timing

QANTAS received three S/360 around 1966, which may have been the first such machines in Australia. The PMG (now Telstra) changed their Telex network from manual switching to Ericsson Crossbar during the 1960s. As far as I can confirm there were 7 multiplexed incoming lines from the USA, 5 from London and perhaps 20 to 40 (unconfirmed) Telex terminals from other parts of the world. The initial speed was 50 bps which was upgraded to 75 bps. The Honeywell project was to attach to telex lines at 110 bps.

Apparently, at one stage the operators managing the paper tape messages would read the tape address and slot the tape into a large board which designated the next leg for transmission. For Example, a message arriving from London but addressed to Wellington would be inserted to the Wellington slot for transmission. By late 1969 messages were spliced onto the end of the appropriate outward going tape and dumped into large bin. For some reason the tape fed out of the bin and never seemed to tangle.

My colleague and I visited the Honeywell plant in Massachusetts,USA during January 1970. I don't recall when the equipment arrived in Australia or when it was installed in QANTAS, which was located near the present site of the Wentworth Hotel, Sydney. We did have a test setup in a facility in Botany Bay where an IBM S/360 was installed along with some Honeywell computers. An H632 was resident at the Honeywell head office in Waterloo for a while.  QANTAS had significant expertise in software development and as far as I recall also developed the software for the H632s.

I don't know when the system became operational, but would very much appreciate any comments that could update my recollections.

A similar configuration to the QANTAS project with an H632 at each end. The two middle cabinets showing the Supervisory Panel, contain common equiment and external interface adaptors. 

The QANTAS machines were only assembled into the final configuration in Australia in 1970.