Re: inserting many records with ADO is very slow

wario_spam_at_insightbb.com (wario) wrote in message news:<962c6d85.0409151034.746db556_at_posting.google.com>...
> Your example would not be the optimal method to perform bulk
> update/insert into a database.
>
> 1. Create a pl/sql block to perform changes.
>
> begin
> execute immediate 'DROP TABLE SCOTT.TEST';
> execute immediate 'CREATE TABLE SCOTT.TEST (NAME
> VARCHAR2(750))';
>
> insert into scott.test (name)
> select name from table_a;
>
> commit;
> end;
>
>
> 2. Better yet. Use SQL Loader to insert all your data into a table
> used for transactions. Then use MERGE, INSERT or UPDATE to
> update/insert changes to main table.

The script is only for testing purposes. The real application generates these records at runtime, so there's no source table i can copy the records from. Tonight i tested it with client side cursors an that seems to solve my problems:

  1000 records 2 s

  2000 records:    3 s
  4000 records:    7 s
  8000 records:   15 s
 16000 records:   31 s
 32000 records:   62 s (server side: 326 s)
 64000 records:  128 s (server side: 2268 s)
128000 records:  249 s (server side: 17533 s)
256000 records:  676 s