Split records into Multiple csv files using a Threshold percentage

Question

Hi Gurus,
I have a requirement to split the data into two csv file from a table using a threshold value(in Percentage) .
Assume that If my source select query of interface fetches 2000 records , I will provide a threshold value like 20%.
I need to generate a csv1 with 400 records(20% of 2000) and the rest of the records into another csv2.
For implementing this I am trying to use the following process.
1) Create a procedure with the select query to get the count of records.
Total Records count: select count(1) from source_table <Joins> <Lookups> <Conditions>;
2) Calculate the Record count to first CSV using the threshold_value.
CSV1_Count=Total records count /threshold_value
3) Create a view that fetches the CSV1_Count(400) records for CSV1 as follows.
Create view CSV1_view as select Col1,Col2,Col3 from source_table <Joins> <Lookups> <Conditions>
Where rownum<=CSV1_Count;
4) Generate CSV1 file using View 'CSV1_View'
5) Generate CSV2 File using the Interface with same select statement (with columns ) to generate a CSV.
select Col1,Col2,Col3 from source_table ST <Joins> <Lookups> <Conditions>
Left outer join (Select Col1 from CSV1_View ) CS on CS.Col1=ST.Col1 where CS.Col1 is null;
Which gives the Total records minus the CS1_View records.
The above process seems a bit complex and very simple . If any changes in my Interface I also need to change the procedure (counts the no:of records).
Please provide your comments and feedback about this and looking for your inputs for any new simple approach or fine tune the above approach.
Thanks,
Arjun

Answer 1

Arjun,
This are my thoughts and Lets do it in 3 Steps
Step 1.  ODI Procedure
Drop table Temp_20 ;
Create table Temp_20 as select * from table where rownum < ( SELECT TRUNC( COUNT(1) /5) FROM TABLE ) .
[ ** This way iam fetching approx 20% of the table data and loading into Temp table . 1/5 th is 20%  so i am dividing count by 5
I don't believe View will help you especially with RowNum as if you run the same query with rownum < N the rows order might differ . so Temp table is great ]
Step 2 .  Use OdiSqlUnload  with select columns  from temp_20
Step 3 . Use again OdiSqlUnload  with  select columns from table where  ( uk keys ) not in ( selecy uk_keys from temp_20)
[** this way you can pick the remaining 80% ** and the data will be not repeat itself across 20% and 80% , as might happen with view ]
what do you think ?