Legacy - Re: [LUG] Linux consoles and proxies

[Simon Waters <Simon at wretched.demon.co.uk>]

John Horne wrote:
> 
> Oh yeah...that makes me feel so much younger :-( Funniest thing with punch
> cards was that it was so easy to read someones password - it was printed
> along the top (usually)! :-) Duh!

Well I punched me own cards for debugging as a school kid,
and having got myself a glorious colour computer in 1982,
figured by the time I got a job such things would be ancient
history.

However when I joined the UK Met Office they still had
systems running on paper tape (Not many) in 1991. Guess it
is all a legacy thing, imagine if like the Met Office you
rented time on the first commercial machine in the UK,
bought by Tate and Lyle, and collected IT equipment ever
since, then legacy really means something. They had 32K PDPs
still going useful work in 1994, a few years earlier having
installed the world most powerful supercomputer (Well so
Cray claimed) on the floor underneath with 256 Megawords
(64bit words) of high performance memory.

When I left in 1996 people were still writing notes to each
other on the back of old punch cards, and they hadn't used
punch cards extensively (Except perhaps for converting
archived data to new formats) for nearly a decade.

When they moved one of their computer rooms one year, they
removed several tonne of old wiring from the hollow floors,
which no one had dare try and untangle for fear of
disrupting the 24x7x365 process that was keeping the worlds
Meteorological data flowing.

It was an interesting way to learn both the value and cost
of legacy systems, and I've never touched IBM MVS since!
Although I still loved the irony of the 'Introduction to IBM
JCL/370', which says (paraphrasing) "IBM Job Control
Language was invented in 1961, when Kennedy was president,
and before some of you were born <sic>(Actually in that
class, before the instructor was born, let alone the
students). Concepts in modern computing have changed
somewhat since then." That last sentence has to go down as
the all time understatement, although the books were a bit
dog eared by then.

It was an interesting look at how organisations got into
mainframe technology, and found it really hard to get their
processes out of this archaic technology. Don't get me wrong
the IBM compatible<sic> mainframes were real hardware
marvels, but they were crippled by trying to support archaic
languages and  conventions. 

You might have a 200 lines JCL job, that the average Unix
box would do with a three line Makefile. Fortran code might
rely on IBMs bizarre static memory allocation, which could
easily halve the performance of the code when we ran it on
systems where you had a choice in the compiler flags.
Character length limits on various names led to bizarre and
arcane naming conventions.

In each case there were ways out of the maze (Better
scripting languages - REXX), or 'conventions and standards'
to prevent you getting into such a mess (No one wrote new
Fortran that relied on static memory allocation - at least
not wittingly), but alas it was the very legacy they were
living with that had given rise to the wisdom on how to
avoid it in the first place.

Those who haven't been to the heart of large corporate IT
systems should be warned that many of the companies I deal
with are still trying to get rid of those lurking
mainframes.

-- 
Simon Waters
Eighth Layer Limited http://www.eighth-layer.com
Tel: +44(0)1395 232769
ICQ: 116952768
--
lug-list - The Mailing List for the Devon & Cornwall LUG
Mail majordomo at lists.termisoc.org with "unsubscribe lug-list" in the
message body to unsubscribe.