Sybrand,
So if I have a table with 20 columns, and I only want 3 of those columns
for a given query, I'm still actually fetching the data from the other 17
columns into buffer cache? Is that what you're saying?
"Sybrand Bakker" <postbus_at_sybrandb.demon.nl> wrote in message
news:06iofu43h4ckkpc15ad8as2nradpcl6j2t_at_4ax.com...
> On Mon, 3 Jun 2002 20:39:32 -0700, "Marc E" <marc_e_at_cablespeed.com>
> wrote:
>
> >Hi all,
> > I have a database theory question. Let's say I have two columns:
> >MyInteger and MyBlob.
> >
> >The table contains 100,000 rows. If I run a select statement against just
> >the MyInteger column, with no reference to any other table, do the
contents
> >of the MyBlob column affect the performance of my query?
> >
> >Put another way, if I have 2 MB of MyInteger Data, and 1000 GB of MyBlob
> >data, and I just want to get at some of the MyInteger data, does that
1000
> >GB of MyBlob data have any impact on my query?
> >
> >Thanks for input.
> >
> >Marc
> >
>
> If you select one column of a record, the whole record is fetched in
> buffer cache. In your case it depends whether the BLOBs are stored
> inline or in a separate blob segment.
> However, I have been working with an application where there was a
> long in a table. The long was needed in only one program, but
> logically belonged to that table. The long was also on average 40k,
> the rest a few measly bytes. Now you guess what happened.
> We created a separate table with the primary key of the original table
> and the long.
>
> Hth
>
> Sybrand Bakker, Senior Oracle DBA
>
> To reply remove -verwijderdit from my e-mail address
Received on Tue Jun 04 2002 - 23:47:05 CDT
