Performanance issue

Question

Hi,
I have a performance issue in my package.
I am getting "comma seperated" records(apprxly 5000) from front end to my package and i am inserting the records into the table by calling
the table package.It is taking 2 mins to load 1000 records.But i should have to load 1000 recs in 1 min.
following are the steps that i am using in my pacakage
<pre>
for i in 1..No_of_recs(say 5000)
1)Calling another function called "subset" to split the comma seperated record(everey record has 10 values,so i have to call this function 10 times for one loop)
2)calling table package to insert splitted values into the table .
end loop.
</pre>
Inside the table i am simply using the insert statement followed by commit statement.
Are there any tips to increase the speed of data insert?
cheers
RRK

Answer 1

As was already suggested, twice, the real solution is to create an external table over you input file and then do the whole thing as a single statement
INSERT INTO table
SELECT * FROM external_tableHowever, if you are unable, or unwilling to do that, then there are a couple of things you can try.
Do not commit after each insert statement, commit once after the entire file has been processed.
Re-write your parsing function so that it splits out all of the column in one shot. If you have 5,000 lines in the file with 10 columns each you are calling that function 50,000 times. Even if that function returns in one thousandth of a second, it is still taking 50 seconds just to call that function.
Consider building a table of records as you parse the file, and then doing a FORALL bulk insert every 100 - 1000 rows, depending on how large each record is.
But, the real solution is still to create an external table over you input file and then do the whole thing as a single statement
INSERT INTO table
SELECT * FROM external_table John