Re: Chained Rows

Daniel Morgan <damorgan_at_x.washington.edu> wrote in message news:<1078266089.98353_at_yasure>...
> Oliver Stratmann wrote:
> > Hello All,
> >
> > we have a table which has about 600.000 rows and the size of about 700
> > MByte.
> > This table was initialized first with not all columns being filled and
> > afterwards calculation-results
> > were added into empty columns.
> > Though before the table was reorganized by a MOVE and the PCTFREE-parameter
> > was set to 40
> > after the calculation 100% of the rows seem to be chained (result of
> > "ANALYZE TABLE T LIST CHAINED ROWS IN CHAINED_ROWS").
> > I read a few FAQs and Docs about this problem but probably i've misread them
> > or overread the important parts of them.
> > I couldn't find an answer.
> > Are there any hints you could give?
> > Thanks in advance!
> >
> > Bye!
> > Oli
> >
> > PS:
> > The Storage-Clause in more detail:
> > PCTFREE 40
> > PCTUSED 40
> > INITRANS 1
> > MAXTRANS 255
> > TABLESPACE tbspace_name
> > STORAGE (
> > INITIAL 314572800
> > NEXT 104857600
> > PCTINCREASE 0
> > MINEXTENTS 1
> > MAXEXTENTS 2000000
>
> For now ... rebuild the table. Then recalculate PCTFREE and PCTUSED
> so that they work. I can't understand why you would use 40 for both
> parameters.
>
> For more information on them go to:
> http://www.psoug.org/reference/tables.html

Isn't the explanation of PCTFREE wrong on that PSOUG page? It states:

"Tables on which there are no updates should have PCTFREE set to 0. The default value of 10 leaves 90% of each block empty."

AFAIK, PCTFREE of 10 will reserve 10% of the space for updates, not leave 90% empty.
This algorithm is used only for INSERTs- A value of 10 means that an INSERT will use that block only if it will leave _at least_ 10% of that block free.
Updates can always use up to 100% of a block. Received on Wed Mar 03 2004 - 08:10:15 CST