I’ve actually loaded AIX 1.3 for the PS/2 on a model 80 decked out with 16MB of RAM, and an Ethernet board, and honestly It was ‘yet another SYSV’ and it didn’t feel like it had much in common with AIX 3.1 on the RS/6000. As always by the time I had gotten all the needed bits, Linux was a thing, running Unix on a 386 with ESDI disks felt horribly slow, and Linux had much better support for stuff. Although if I had the machine when it was the thing to do it’d have been awesome. Not that I’d probably ever get access to a 370, let alone AIX for the 370 + those i860 boards. The 80 was a great machine, just too bad that I’m sure they called it the model 80 as it must have weighed 80Kg, and I couldn’t take it with me when I left North America. Back when I used to work for a certain bank that loved mainframes, they would always harp on about how Unix & C were not only untested, but simply not ready for an environment like the S/390. As far as they were concerned if it didn’t boot on a mainframe it wasn’t “production grade”. Oddly enough we also ran things like Novel Netware for VMS (Wish I could have imaged those tapes...), although they were kind of OK with AIX on the RS/6000, but never for anything ‘real’. It wasn’t until after I had left that I found out that there actually was mainframe UNIX, and if it would work with our ‘awesome’ global SNA network and zillions of terminals it would have been all the better. Sent from Mail for Windows 10 From: Clem Cole Sent: Thursday, 2 March 2017 11:16 AM To: Jason Stevens Cc: TUHS main list; Ronald Natalie Subject: Re: [TUHS] AIX/370 Questions On Wed, Mar 1, 2017 at 9:13 PM, Jason Stevens <jsteve(a)superglobalmegacorp.com> wrote: Slightly off or on topic, but since you seem to know, and I've never seen aix 370 in the eild, did it require VM? It could boot on raw HW.​   Did it take advantage of SNA, and allow front ends, along with SNA gateways and 3270's? ​Not sure how to answer this.​  It was an IBM product and could be used with a lot of other IBM's products.  Generally speaking it was aimed at the Educational market, although there were some commercial customers, for instance Intel was reputed to do a lot of the 486 simulation on a TCF cluster (I don't know that for sure, that was before I worked for Intel).   Or was it more of a hosted TCP/IP accessable system? ​Clearly, if you had a PS/2 in the cluster, that was your access point.   I think it was all mixed up in the politics of the day at IBM between Enterprise, Workstations, and Entry systems.  TCP/IP and Ethernets were not something IBM wanted to use naturally.    But the Educational market did use it and certainly some folks at IBM saw the value. UNIX was needed for the Education market as was TCP/IP so that going to be the pointed head of the stick.

On Wed, Mar 1, 2017 at 10:51 PM, <jsteve(a)superglobalmegacorp.com> wrote: ​Interesting characterization.... It had all of the primary BSDism in there. As I said it was intended for the Education market., and it would have been lost without it. I never saw anything from a BSD that did not pretty much just recompile on AIX/386 - but I admit I never pushed it very hard, since it was not a primary system for me. ​It's too bad you missed one of the coolest tricks that TCF would allow you to do​. If you had a running TCF node (370 or 386) with a full install on it, all you needed to install another node was the boot floppy. The OS would load into memory from the boot floppy, send out a couple of packets on the network (never touching its disk) and the node would join the cluster and immediately be a part of the system. If the was disk was empty, it would start to be populated in the background using replication, if it had data, then the rebuild protocol started -- al behind the scene. It meant a node boot was really the time to perform the cluster management join protocol, which on TCF was fairly fast because it was bounded to 32 nodes. Later TNC clusters the CMS was a lot more complicated. All very, very cool. I do miss some of those ideas. Clem