From coff at tuhs.org Fri Nov 13 00:33:34 2026 From: coff at tuhs.org (Charles H Sauer (he/him) via COFF) Date: Thu, 12 Nov 2026 08:33:34 -0600 Subject: [COFF] [TUHS] Re: AS/400 and TUI worlds: were they (very) isolated from UNIX? In-Reply-To: <1ED75CE0-9E7F-4F22-9335-AFA2E055331E@canb.auug.org.au> References: <20261112021405.59F161282D0EF@ary.qy> <1ED75CE0-9E7F-4F22-9335-AFA2E055331E@canb.auug.org.au> Message-ID: <2cdf7350-eded-4689-9b8c-611c445dc6b5@technologists.com> Discussion of AS/400 probably belongs in COFF. Contemplation of AS/400 should begin with its initial instantiation, the System/38 -- see https://bitsavers.org/pdf/ibm/system38/IBM_System_38_Technical_Developments_Dec1978.pdf. IBM largely precluded employees' involvement with Unix until well after the System/38 was released, so it wasn't until many years later that there was significant influence of Unix on the AS/400. Glenn Henry was the principal person behind the System/38. IBM would not have succeeded in the Unix business without Glenn's impetus after moving from Rochester to Austin, after finishing his work with the System/38. Glenn instigated and led the development of what became the RT/PC and AIX. Without the RT and associated AIX experience, IBM likely would not have developed (or succeeded with) the RS/6000, PowerPC et al Glenn has been working on a memoir "Computers I Have Known" -- see https://www.glennsmuseum.com/intro.pdf and Glenn's LinkedIn posts. Though the planned chapter on the 38 is not yet visible to my knowledge, at https://www.linkedin.com/feed/update/urn:li:activity:7510106472228265984/, Glenn says "Keeping this project alive and getting it shipped while retaining its advanced and unique technical features (high-level object-oriented machine interface, single-level storage with 48b addressing, integrated relational database, etc.) in spite of heavy resistance from the “establishment” was the hardest job I have ever done, and the one I am still most proud of." I'm also working on a memoir at https://technologists.com/GMT2M/. It is relatively complete through 1995 (with outline and snippets for subsequent years) and discusses working for Glenn at IBM and Dell. Charlie On 11/11/2026 10:40 PM, steve jenkin via TUHS wrote: > >> On 12 Nov 2026, at 13:14, John Levine via TUHS wrote: >> >> It appears that Tom Lyon via TUHS > said: >>> ... > >>> Given it's origins in the mid 70s, I doubt any UNIX influence. But - >>> smells strongly of Multics to me - single huge address space for everything. >> >> Not really. Multics was segmented, i has a large flat address space full of objects. >> I gather that S/38 and AS/400 were salvaged from the failed Future Systems project >> that was supposed to repalce S/360. >> >> And there's PASE, a full Unix environment on top of i. > > S/38 & AS/400 used some Multics concepts quite uniquely. > > I did a summer job on System/34 once [ popular small accounting machine, later S/36 ] > then had to deal with System/38 a few times. > Each time I had to use a variant of RPG. > > System/38 were very popular in business and Libraries (!) - they came with a native database. > And didn't need a Systems Admin, just a somewhat trained user. > > S/38 had two distinctive features: > > - loadable instruction sets per Application [ microcode sets ] > - very large (48-bit) virtual memory. > > Everything, including files, was an object in Virtual Memory. > A few times I was around S/38 or AS/400 the 48-bit VM space was exhausted > and they required reboot. The VM Segments were never reused. > > The System/36 and System/38 lines were merged into the AS/400. > [ Then the AS/400 evolved... Didn't track i & p series ] > > AS/400 Objects & Single-level Store > > > S/38 > > > The AS/400 kept the native database and 'everything an object', > though seemed to have moved to a 64-bit Virtual Address. > > Another S/38 or AS/400 'feature' was a dealing with problems during a scheduled power-down then restart. > > A datacentre 3-phase was upgraded and two-phases were swapped, > which caused some 3-phase motors to turn backwards [ killed some Univac disk drives ] > with an emergency power-down of *everything*, [ "hit the big red button" ] > including the AS/400, while it was restarting - before it had fully counted it's fingers & toes. > > The AS/400 decided the database indexes must be corrupt due to the power-fail, > so automatically rebuilt all indexes when power was eventually restored. > > The power was restored by 5PM on a Sunday, the indexes took until close of business Wednesday to rebuild. > In the interim, the whole of the Library system was unavailable - they couldn't even login I believe. > [ A public library system that did a lot more than the catalogue ] > > I had a clutch of SUN systems caught up in this power outage. > Don't recall we lost any systems or components. Our fsck's didn't take three days to finish :) > > There was another time in the same datacentre, which always ran off a UPS, never direct mains power or diesels, > where a sparkie ignored a prominent notice in the distribution board > and disconnected a ground wire, which the UPS relied upon to detect current leakage. > > It wasn't a clean 'power off', the UPS went up and down multiple times over a few seconds. > That blew up multiple bits of kit, including their main CISCO router - unsure if duplicated or not. > They had a 3-hr response service contract with CISCO, > who got the part down from Sydney and installed in good time - kudos to them. > > The AS/400 survived that debacle from memory. A bunch of other systems & networking kit didn't fare well. > > > -- > Steve Jenkin, IT Systems and Design > 0412 786 915 (+61 412 786 915) > PO Box 38, Kippax ACT 2615, AUSTRALIA > > mailto:sjenkin at canb.auug.org.au http://members.tip.net.au/~sjenkin -- voice: +1.512.784.7526 e-mail: sauer at technologists.com fax: +1.512.346.5240 Web: https://technologists.com/sauer/ Facebook/Google/LinkedIn/mas.to: CharlesHSauer