I am using Oracle 81630 on Solaris 7 and I have a table which contains
a
LOB (CLOB actually). The DDL statement is as follows ...
CREATE TABLE B_STATEMENT_TEXT_ORA (stmtnum NUMBER(10) NOT NULL,
statement_content CLOB)
STORAGE (INITIAL 100M NEXT 100M PCTINCREASE 0)
TABLESPACE energy_proc_data_100M_02
LOB (statement_content) STORE AS b_statement_text_ora_lob
(TABLESPACE energy_proc_lob_100M
DISABLE STORAGE IN ROW
STORAGE (INITIAL 100M NEXT 100M PCTINCREASE 0)
CHUNK 8192
PCTVERSION 10
);
The details from user_lobs are ...
SEGMENT_NAME
TABLESPACE_NAME
SEGMENT_TYPE
BYTES EXTENTS
------------------------------ ------------------------------ ------------------
---------- ----------
B_STATEMENT_TEXT_ORA
ENERGY_PROC_DATA_100M_02 TABLE
104857600 1
SYS_IL0000046179C00002$$ ENERGY_PROC_LOB_100M
LOBINDEX
104857600 1
B_STATEMENT_TEXT_ORA_LOB ENERGY_PROC_LOB_100M
LOBSEGMENT 104857600
1
The DML operations on the table include SELECT, INSERT and DELETE.
There is no UPDATE activity.
Typically the data for each CLOB entry is larger than 4K so I am
specifying that it is stored in the lobsegment and not the table.
My question is one related to data access for the table, lobsegment
and
lobindex. Would there be any I/O performance benefit in having
the
table, lobsegment and lobindex located in different tablespaces (which
would be located on different physical drives)?
As you can see I can split the table and lobsegment/lobindex into
separate tablespaces. I have tried to split the lobsegment and
lobindex
up into separate tablespaces but according to the 816 documentation
this
is no longer supported under 8i (depracation of the LOB_index_clause).
According to Metalink the lobsegment and lobindex are co-located in
8i
for a specific reason - but I cannot find what that is!
Also, is there any way to specify a name for the lobindex, rather than
rely upon the system generated name?
Thanks
Glen
-- Glen Mitchell NZ Phone: +64 9 3730400 Energy Research Lab URL: http://www.peace.com Peace Software Email: [EMAIL PROTECTED]
