What ever happened to Tandem and NonStop OS 2119

when i was doing page replacement algorithms as undergraduate in the 60s

one of the principles was to perform overhead directly related to some activity and-or resource constraint.

at the time, i noticed that the version of tss-360 ... when activating a tasked characterized as "interactive" one first attempt to move all the related virtual pages from 2311 (moveable arm) disk to the 2301 (fixed head) drum before actually starting the task running ... and then when the task suspended ... move all the related pages off the 2301 drum back to the 2311 ... even when there was no contention for 2301 space.

part of this was that i was well into dynamic adaptive algorithms at the time ... not only for virtual memory scheduling resources but also for dispatching and managing other resources. this is in part, where the "wheeler" and "fair-share" scheduler originated

... although the scheduler wasn't specifically fair-share ... it had resource policies ... where a possible default policy was fair-share.

in the very late 70s ... I got contacted by somebody from POK that had made a major change in MVS to not arbritrary flush pages from real memory back to disk when task quiesced-suspended in cases when there wasn't contention for real memory. He had gotten a large corporate award for the change ... and wanted to get another award for doing the same thing to vm370. I commented that I had never, not done it that way ... dating back to when I first started ten years earlier. Furthermore I had tried to convince the people that were responsible for the SVS (and then MVS) implementation to not do it that way back in the early 70s ... and they insisted on doing it anyway.

i may have even commented that in addition to given an award to the person that fixed it ... that possibly at least the amount of the award should be deducted from the compensation of the people that had insisted on doing it wrong in the first place.

--