The question was “what exactly was/is a mainframe?”

So – in my opinion, “back in the day” – BEFORE windows/Linux/unix/intel/dell (from 1950s thru, let’s say: 1985-1990 – before Unix/windows NT servers came into existence), and take this (at first) with a grain of salt:

A mainframe was not ‘commodity’ hardware; each computer manufacturer (before AND during the advent of the seven dwarfs / BUNCH), had their own proprietary hardware AND OS software. Once you bought IBM, or Burroughs, or NCR, CDC, Sperry/univac, Honeywell, you were ‘stuck’ with them since each manufacturer – (while most all of them did COBOL, FORTRAN, etc.) had proprietary ‘stuff’ – moving from one manufacturer to another was very hard to do: their customers (us) had a significant investment in training their programmers, mainframe support people on each manufacturer’s set of tools (compilers, editors, “JCL”, schedulers, backup software, etc.); AND, their peripherals (tape drives, printers, disk drives) definitely were not plug compatible with any of the other manufacturers. [this excludes the on-purpose ‘plug’ compatibility between IBM, Amdahl, Fujitsu]

Notice I did not (nor, did anyone else at the time) lump DEC, or HP with the seven dwarves/BUNCH – and while it was true these two companies also had proprietary HW and SW, they were significantly less costly to buy, on a factor of 10, or even 100. For example, a “mainframe” could set a company back $1 mill to $10 mil. On the other hand, both HP and Dec were selling their stuff from $10k to 100k (let’s not get collective our shorts in a knot, these are just examples and just trying to make a point regarding the significant pricing difference). Therefore, these weren’t called mainframes, but a new class was ‘invented’ for them: mini-computers, and/or mid-range (maybe).

The thing I’m not clear about is how ibm Os400/s-36/s-38 class machines were classified – anyone heard reference to them as mainframes, or maybe just mid-range?? These, I think (and maybe a portion of the HPs, or DEC VAX’s) did not require raised floor, and/or separate cooling systems, etc. That was another way to differentiate between what a mainframe is/was, and what was not.

There is no one sentence description of what a mainframe is; it was a set of things, when grouped together (as noted above) – which distinguishes them from ‘non-mainframes’.

Lastly, I’d be significantly remiss if I didn’t mention processing power – even when Unix/linux/winNT servers started becoming the “norm”, these new server types could not process the same amount of transactions (online/database/batch, etc.) as mainframes did at the time; AND, they were not as robust – in terms of database or online transaction processing – i.e., when systems crashed or disks crashed, and data had to be reloaded from tape, or databases had to ‘recover’ automatically, etc. [this was BEFORE RAID technology].

I’m sure everyone has an opinion on what a mainframe is/was; no one person is right, but no one is wrong either with their description of mainframe.

😱😉🤓💻