Hmmm, but if I try to select * from table where there are 1.5m records, the
select alone can take a very long time. When processing finally starts, it
would appear that the select is complete and processing takes a very long
time too. There is what seems to be endless disk activity while all this is
going on.
OTOH, if I structure the select as I suggest, I can turn an "endless"
process into something which feels quite light and can take subminute
durations compared with several hour durations.

Alan
-----Original Message-----
From: Ann W. Harrison [mailto:aharrison@...]
Sent: Tuesday, 12 November 2002 10:23
To: ib-support@yahoogroups.com; ib-support@yahoogroups.com
Subject: RE: [ib-support] Database won't close.


At 09:52 AM 11/12/2002 +1100, Alan McDonald wrote:

>if you pull [all 1.5 million records] in as one select, then I don't
>call it a simple process. You are pulling the entire database off
>the disk into memory which means it's going back on to the disk as disk
>buffer, pretty useless exercise...

No, not exactly. At least that's not what the database does. The
database keeps some finite number of pages buffers in its cache.
As it scans through the data, it reads pages into buffers. When
all buffers are full, it releases and reuses the least recently
used buffer. The operation, if I understood, was read only for
the database, so the least recently used page is just overwritten
in memory - not flushed to disk.

>I
>would expect the database to more than tripl in size under this procedure
>since there would be so much garbage... maybe you are running out of
>diskspace too :-)

In theory, there was no update thus no garbage and no growth...



Regards,

Ann
www.ibphoenix.com
We have answers.



To unsubscribe from this group, send an email to:
ib-support-unsubscribe@egroups.com



Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.



[Non-text portions of this message have been removed]