Re: CHAIN_CNT > 0. Why?

Syltrem wrote:

> I have some tables for which CHAIN_CNT (after ANALYZE) is > 0;
>
> OWNER TABLE_NAME AVG_ROW_LEN NUM_ROWS CHAIN_CNT
> NUM_COLS
> --------------- -------------------- ----------- ---------- ---------- -----
> ---
> FIN BIN_LOCATIONS 191 101028 689 40
> FIN DESPATCH_HEADERS 514 215763 7800 58
> FIN GEM_FACILITIES 375 8439 34 13
> FIN GL_ACCOUNT_BALANCES 345 24358 21 55
> FIN SALES_CREDIT_NOTES 456 13031 102 84
> FIN SYS_PRODUCT_LANGUAGE 57 48343 141 6
> IPS IFA_CUST_PROFILE 266 1160 19
> 46
>
> 7 rows selected.
>
> Why is it so?
> - All character columns are fixed length so that can't change with an update
> - Number of columns range from 6 to 84
> - Row lengths is a lot less than my 8k Oracle blocks
>
> Are number columns like NUMBER(20,2) stored in such a way that the space
> occupied will vary with the value therein?
>
> If not, what should I watch for, as a cause for these rows to be
> migrated/chained ?
> Could not find much more in the docs so far.
>
> Thanks
>
> Pls don't flame about Oracle version - I'm migrating to 817 or 9.2 as soon
> as my software vendor allows.
>
> --
>
> OpenVMS 7.3-1, Oracle 8.1.6.0
>
> Syltrem
> http://pages.infinit.net/syltrem (OpenVMS related web site - en français)
> To reply to myself directly, remove zulu from my address

Numbers are variable length and you should not use CHARs as they waste space and make string comparisons far more problematic.

--
Daniel Morgan
http://www.outreach.washington.edu/extinfo/certprog/oad/oad_crs.asp
damorgan_at_x.washington.edu
(replace 'x' with a 'u' to reply)