20 Feb
2012
20 Feb
'12
5:34 a.m.
On 20 Feb 2012 at 15:34, Nigel Williams wrote:
The term 'mainframe' helped to categorize the
physical aspect of
machines from the 1950s through 1980s , but since then it has
diminished as a differentiator.
I think we're getting tangled up in what's sufficient to call
something a mainframe (e.g. it weighs 20 tons and requires at least
50 tons of cooling to operate) and what's necessary ( ...???...) for
something to be called mainframe.
Can anyone fill in the ??? is an absolute and lucid way, such that
any box, regardless of appearance must pass the test in question to
qualify as a "mainframe"?
I don't think so.
I think that "what it's used for" isn't a valid criterion. I've
used
many pieces of iron that were used for nothing more than compiling
and running simple programs their entire service life. Some
mainframes were little more than I/O front-ends for other mainframes.
Hence my paraphrase of the famous Justice Potter Stewart quote and
an oblique reference to the elephant test.
--Chuck