Splitting strings contained in the database isn't really something that databases were designed for.
There's different options here:
- restructure your data to make strings like
10,244,566
- three columns, if it's always three values, or
- a n-to-m mapping with a second table
- program the splitting in whatever language you're using to talk to postgresql
- hack together a relatively hard to debug, relatively slow, relatively "far away from what you want" extractor in postgresql:
select substring(the_column from ',\d+,') from the_table
I strongly recommend doing option 1.1. if possible, or option 1.2. if the string might not always represent three values.
As it is now, you're abusing your database's string data type to store three integers, which is a waste of space, slow, error-prone, an anti-pattern and hard to maintain.
EDIT: I have to take back a bit about you abusing a database: whoever is pushing that string into postgresql is abusing the database, so that you now have to deal with it :(
EDIT your clarification is critical:
I am proccessing the string in postgresql function. My function is connecting to another component (lets say paypal) and reciveing a data burst as STRING. I need to convert this String for items to store in my PostgreSQL DB. So I catualy saves the 2nd parmeter as integer
Awesome, so just use the programming language you're using to get your string from let's say paypal and split it to three parts there. There's no reason postgreSQL would do that better than you could, and you would have the conceptual benefit of transporting well-formed data to your database.