When I joined, they had a 360/25 with 48K of main memory, three or four 2314 disk drives (one non-IBM), a couple of tape drives and a dozen or two local 2260 terminals used for order entry. The operating system was DOS/360.
At some time, the 360/25 was replaced by a 370/125 and DOS/VS.
When I left, they had a 370/135 with 3330 disk drives and a couple of hundred remote 3270s in the field (the warehouses). The operating system was DOS/VS, the TP Monitor was THIRST (Task Handling with Immediate Response for Terminals), developed and written by Geoff Brown, and direct access files were mainly ISAM (I don't think they went VSAM while I was there).
They still called their customers most every day but instead of writing the order down on paper, the order clerks entered the order in the computer while the customer was on the phone. The system produced a combined pick-list (for the warehouse operations) and invoice (for the customer) and had a crude out of stock mechanism. For the early '70s, this was quite an advanced system.
I became a Systems Programmer while we were starting to design the new order entry system, and contributed by writing some of the Systems Software required. Specifically, I designed and wrote:
The Product Identification system had interesting requirements. The order clerk had to identify the desired product within 6 seconds while the customer was on the phone, but we could not rely on standardized product codes. We also had to be flexible enough to use the existing product names and realize that the customer could describe a product in many different ways. I used two techniques to make this a manageable project:
TOSS (Terminal Oriented Spooling System) was a remote spooling system for 3270 type printers. The order system produced a combined pick-list and invoice, which had to be sent to a remote printer in the same warehouse. Each terminal or group of terminals had a default printer. If there were problems with a printer, its work load could be transferred to a different printer.
I was surprised I was allowed to call it TOSS - and yes, it did have an output file called TOSSOF.
In that same vein, THIRST programs (which were pseudo-conversational) all started with a table and
a bit of code that identified where it was in the process and what it should be doing next.
We called the table the Concise Leg Initiation Table.