My program would take IOTRACE tapes as input, to detect the actual access patterns for each segment of each file, and then repositioned the files across the disk farm to enhance throughput based on the file and disk characteristics.
I called it IFARTUFART: Interactive File Analysis, Relocation and Tracing for the Ultimate in Faster Access and Response Times...
Ahhhh....those were the days!
We would have to take the mainframes down at night and there was always one or two users who wouldn't logoff. On the CDC mainframes you could sent a user a message with ":D" in it and they would get automatically logged off.
The CDC mainframes had a CPU that had no I/O capability but read/write memory (Seymour Cray designed them to run circles around everyone else). You had either 10 or 20 peripheral processors (PPs) that talked to the channels where the tape drives, printers, etc were managed. One PP was reserved for monitor and one for the console display driver. The rest were assigned as needed by the OS. I programmed the PPs.
Once I was working on a proposal and the Government asked my boss in a design review how I came up with the Lines of Code estimate for the PP driver. I told her, "Cause I already wrote it." She says, "I can't tell them that!" So she goes into this song and dance about "based on past experience..."
The assembly language on a CDC mainframe was 64 instructions (the first true RISK architecture). Taught myself when I was a computer operator back in '72.