Note in-line
--
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
June 2004 UK - Optimising Oracle Seminar
July 2004 USA West Coast, Optimising Oracle Seminar
August 2004 Charlotte NC, Optimising Oracle Seminar
September 2004 USA East Coast, Optimising Oracle Seminar
September2004 UK - Optimising Oracle Seminar
"MAK" <maks70_at_comcast.net> wrote in message
news:b7178504.0404281447.18f736d_at_posting.google.com...
> OK I did further tests. I got all the reported chained rows from
> temp_s_contact_03 into another table TEMP_S_CONTACT_04. and again all
> the reported chained rows from TEMP_S_CONTACT_04 into
> TEMP_S_CONTACT_05. Here are my results. Chained rows are reducing in
> each iteration.. Can you explain why?
No - not without setting up a couple of tests
and dumping the data to see whether your
version of Oracle behaves differently from
8.1 when handling multi-piece rows
> From what you explained earlier, nos of migrated/chained rows should
> have stayed the same with each iterations. Right!!!
>
Yes - that's why I assumed your problem wasn't
about rows with more than 250 columns when I
read your first post. But as I said in my last post,
maybe Oracle has changed some of the ways it
can handle such rows.
> TABLE_NAME CHAIN_ROW_COUNT
> ------------------------------ ---------------
> TEMP_S_CONTACT_03 6056
> TEMP_S_CONTACT_04 79
> TEMP_S_CONTACT_05 1
>
> Here are the nos of blocks for each segments. There were no initial,
> next extent submitted. and tables were created with default storage
> settings in DM tablespace. Here is the nos of blocks (8k) for each
> table.
>
It's a pity you mentioned dictionary managed tablespaces -
I was just beginning to wonder if this was a fringe benefit
of using Auto Segment Space management - which is
supposed to require locally managed tablespaces.
> Table_name Blocks
> -------------- ---------
> TEMP_S_CONTACT_03 49607
> TEMP_S_CONTACT_04 615
> TEMP_S_CONTACT_05 133
>
> Thanks
>
Suggested experiment:
create some 300 column rows that are only 1,000 bytes long.
See if Oracle will allow the two pieces to be in different blocks
See if they are reported as chained when
a) the two pieces are in the same block
b) the two pieces are in different blocks.
There's a recent article about row-chaining here that might
be of interest.
http://www.tlingua.com/articles/rc.html
Received on Thu Apr 29 2004 - 00:26:08 CDT
