Re: database table write on read

Just a thought, wouldn't an ALTER SYSTEM CHECKPOINT force all the dirty blocks to get written? Or is "block cleanout" something different? Wouldn't this be easier than select count(*) from each table?

MotoX wrote:

> Yep, it's 'block cleanout'.
>
> I had the same problem after largm data loads into datawarehouse tables.
> Oracle Tech. Support suggest that once these big transactions have run you
> do a 'select count(*)' from each table.
>
> MotoX.
>
> Duane Tiemann wrote in message
> <6p5hn9$gh8$1_at_newssvr04-int.news.prodigy.com>...
> >We have observed some file write activty by Oracle which we don't
> >understand. We've replicated the situation on a test system as follows:
> >
> >We insert a few rows which have a long raw column with 64000 bytes in it.
> >After waiting several minutes (in which the rows seem to be successfully
> >logged and written to the database) we retrieve any value from one of the
> >rows.
> >
> >The retrieval results in write activity to the table. This is documented
> by
> >"select * from v$filestat" as well as AIX's iostat; The number of blocks
> >written is the same as the number of database blocks included in all the
> >rows inserted.
> >
> >Can anyone tell us why this write activity occurs?
> >
> >BTW, the 'extra' writing does not occur if we place 6K or less in the long
> >raw column. It also does not occur on subsequent retievals -- only the
> >first.
> >
> >More details:
> >
> > Oracle Version 7.3.3
> > AIX Version 4.2.1
> > Database block size 16384
> >
> >Thanks,
> >
> >Duane Tiemann
> >Prodigy Services Co.
> >duanetiemann_at_prodigy.net
> >
> >
> >
Received on Thu Jul 23 1998 - 20:07:00 CDT