Re: Hashing for DISTINCT or GROUP BY in SQL
Date: Thu, 14 Oct 2010 08:24:43 -0700 (PDT)
Message-ID: <57b081be-5d87-4bf1-891f-0fc98756c159_at_x23g2000vba.googlegroups.com>
On Oct 12, 1:08 pm, Roy Hann <specia..._at_processed.almost.meat> wrote:
> -CELKO- wrote:
> > How many kids
> > today have ever seen a polyphase merge sort on tape drives?
>
> I turned 50 last birthday and I've never seen one, and I live in
> England.
>
> --
> Roy
After the late 1960's at least in the IBM part of the western dp world
(excluding, for example, India), most installations had disks to use
for 'sortwk...' (aka sort work files). Tape sorts were pretty
hilarious to watch, usually two or three tape drives were used as
'work area'. In some installations, close to 50% of those runs had to
be re-started because of tape head dirt or other malfunctions. It was
important to use fairly new tape and to clean the heads first. From
moment to moment, the drives would switch between 'read' and 'read
backwards' which played hob with the tape motors and drive belts and
the thin 1/2" tape. Movie crews liked them for the appearance of
meaningful action. I'd say they were an order-of-magnitude slower
than disk-based sorts. Programming in those days had a high ratio of
manual labour.
81b39258-7323ab72-68a05e2552
Received on Thu Oct 14 2010 - 17:24:43 CEST