| Faster Context Index creation!! [message #135168] |
Tue, 30 August 2005 04:20  |
ankurgodambe
Messages: 45
Registered: March 2005
|
Member |
|
|
Hi,
I had post this question on Text forum but didnt get a reply. I think this is the right place to put it.
My problem is that we have monthly process of rebuilding the context index on a table ( varchar column). I cannot stop the process of rebuilding as it was suggested by Oracle. This process takes about 8hrs. When I created it the last time I increased the sort_area_size parameter for the session to 500M,increased sort_multiblock_read_count and db_file_multiblock_read_count depending upon the OS limitations,and the index was created in 4 hrs. But this is just DBA trick to do things faster and i beleive it can be done more faster.
Can anyone suggest me what are the ways of speeding up the index creation process from the CONTEXT perspective
,like increasing default memory with Ctx_Adm.Set_Parameter ( 'DEFAULT_INDEX_MEMORY', '500M'); .
Also having default_index_memory and sort_area_size as 500, will this take 1000M of memory during the index creation?
Also i read somewhere that before creating index truncating the table DR$INDEX_ERROR will help speeding index creation. Is this right? I think it should not make a difference .
Any suggestion on speeding up the index creation would be helpful. i cannot create the index in parallel as I am running Oracle 8.1.6 and base table is not partitioned.
Thanks.
Ankur
|
|
|
|
|
|
| Re: Faster Context Index creation!! [message #136697 is a reply to message #135168] |
Fri, 09 September 2005 10:10  |
ankurgodambe
Messages: 45
Registered: March 2005
|
Member |
|
|
Hi,
I rebuild it by dropping the index and then creating it again. I do not use CTX_DDL.SYNCHRONIZE and CTX_DDL.OPTIMZE as we have observed that the performance is much better when we drop and create the index than using CTX_DDL.SYNCHRONIZE and CTX_DDL.OPTIMZE .
Regards,
|
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
