Re: Hashing for DISTINCT or GROUP BY in SQL

From: Cimode <cimode_at_hotmail.com>
Date: Fri, 15 Oct 2010 13:26:53 -0700 (PDT)
Message-ID: <3df0993a-9c00-4d33-97d3-317022266159_at_i21g2000yqg.googlegroups.com>


On 15 oct, 19:25, paul c <anonym..._at_not-for-mail.invalid> wrote:
> On 15/10/2010 9:59 AM, -CELKO- wrote:
>
> > The later units had vacuum columns instead mechanical arms to control
> > the tape's momentum. Did you ever see a data cell machine? It looked
> > like a giant version of an amusement park claw machine. It grabbed a
> > cylinder with a ribbon of mag tape, wrapped the tape around a drum,
> > read or wrote it, then put it back in a pigeon hole.
>
> I was given one or should I say I just took it.  I worked for a
> hole-in-the-wall research company and we couldn't afford the
> maintenance.  The parts were all what I called NASA-quality, eg., the 3
> HP three-phase electric motor had a label on it saying 'lubricate every
> 20 years'.  The IBM model number was 2321.  It was ten years old and had
> 1,700 hours on the clock, so the university we got it from obviously had
> problems keeping it running too.
>
> I wanted the beautiful little rotary compressor but was afraid to
> disconnect all the hydraulics because of many labels warning 'danger -
> compressed nitrogen'.  Finally reached a former IBM Customer Engineer
> who had serviced them before getting a desk job.  His first question was
> "where is it?"  When I replied "in the computer room", he laughed and
> told me the main danger was undoing the connections in the wrong order,
> which could result in the large hydraulic fluid reservoir (I forget how
> many gallons but it was about a foot high and a couple of feet across)
> suddenly squirting all that oil in all directions.  So I rolled it out
> to the parking lot (must have weighed close to 1,000 pounds) and took it
> apart there.  Took me a couple of days to completely strip it.
>
> I remember clearly the average access time (what some today call
> latency): 500 ms / half a second (even in those days, a small 2311 or
> 2314 disk drive had an average seek time of less than 30 ms).  I think
> some sites used it for sortwork.  In those days (1970's) all programmers
> had to know how to set up a sort (well, may not the Cobol programmers!).
>   After a while, you'd get to know the error codes by heart.  IBM had
> competitors selling faster sorts, such as Syncsort and CASort.  As I
> recall, the IBM messages started with the code 'ice'.  Syncsort used to
> run mocking ads in the trade press referring to the IBM product as the
> "ICEMAN".  There were a bunch of other IBM utilities many of which were
> really just 'legacy' support for unit-record systems.  But old-timers
> continued to make new 'systems' with them.  There were several whose
> names you couldn't pronounce without spitting, eg., they had too few
> vowels, such as 'iebptpch'.  I used to refer to it as the 'pteradactyl'.
>   Somewhere on the internet there is probably the manual for it, if you
> read it you'll have new respect for the word 'arcane'.
SOC7 !!! Your comments reminded me of what initially triggered my interest in databases in the first place: the proto-index structure logic behind the early VSAM systems. At that time, the initial direct image data files were called clusters. Not much have changed since, regarding database implementations. Received on Fri Oct 15 2010 - 22:26:53 CEST

Original text of this message