[email protected] (Daniel Skwire) writes: > I thought the FAA had special hybrid 6 computer systems, 3 x 2 way MPs?
re: http://www.garlic.com/~lynn/2013o.html#54 Curiosity: TCB mapping macro name - why IKJTCB? http://www.garlic.com/~lynn/2013o.html#55 Curiosity: TCB mapping macro name - why IKJTCB? IBM 9020 http://en.wikipedia.org/wiki/IBM_9020 from above: The IBM 9020A, for example, was based on the System 360/50 and the 9020D used 2 out of 3 or 4 360/65 processors for flight and radar data processing with 2 out of 3 360/50 processors providing input/output capability. ... and The 9020As and 9020Ds were in service in North America until 1989 when they were finally replaced by IBM 3083 BX1 mainframes as part of the FAA's HOST upgrade. ... snip ... the wiki entry also references IBM system journal article from 1967: "An application-oriented multiprocessing system, Part II: Design characteristics of the 9020 system" ... they have been moved behind paywall at IEEE. couple trivia. originally there wasn't any plans for 3083 ... just 3081 as dyadic and pair of 3081s for 4-way 3084. big problem was that ACP/TPF (airline control program renamed transaction processing facility) didn't have multiprocessor support. http://en.wikipedia.org/wiki/Transaction_Processing_Facility there initially was some very unnatural things done to vm370 for running on 3081 multiprocessor done to improve throughput of TPF running in virtual machine ... that turned out to degrade the multiprocessor throughput of almost every other customer. eventually there was decision to remove one of the processors in the 3081 cabinet to come up with 3083. a problem was that everything was wired for processor0 (the non-removed processor) was at the top of the cabinet, just removing processor1 in the middle of the cabinet left the box dangerously top heavy. there was lots of concern that all the TPF customers would all move to clone processor vendors ... which had faster, more modern single processor machines. the other issue was the significant competitive issues with the 308x technology compared to clone competition ... discussed in some detail here: http://www.jfsowa.com/computer/memo125.htm both 3033 & 3081 were mad rush efforts in the wake of the failure of FS project ... using some over technology warmed over from FS http://www.garlic.com/~lynn/submain.html#futuresys TPF eventually did come up with multiprocessor support ... didn't take quiet as long as it took for CICS to come up with smp support (2004). .. the scientific center ... some past posts http://www.garlic.com/~lynn/subtopic.html#545tech started its virtual machine / virtual memory effort before standard 360/67 was available ... so they first tried to get a 360/50 to do their own hardware modifications to support virtual memory ... however all the spare 360/50s were going to FAA ... so they had to settle for 360/40. comments was that they were glad they got 360/40 since the hardware changes for virtual memory support was much simpler than what they would have had to do for 360/50. thus was born original cp40/cms ... some description http://www.garlic.com/~lynn/cp40seas1982.txt cp40/cms later morphs into cp67/cms ... which then morphs into vm370. .. there were a number of FAA modernization efforts ... several of them not making to fruition. when we were doing ha/cmp product http://www.garlic.com/~lynn/subtopic.html#hacmp we got pulled into doing some project reviews. one such involved triple-redundant hardware and the people writing the application software were told that they didn't have to program for errors or failures ... since the system would mask all faults to the application level. the problem was that there are a number of (flight control) business process level failures that have to programmed for. -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
