This is a response, not a review
Sankalp, October 23, 2017 - 9:12 am UTC
Sorry, I could not find any option to respond, only got a review option.
You understood it correctly. There is already parallelism given in my original code. My question is how can I do it faster.
With a parallelism of 15, it is taking 3.5 hours to work on 7mn data. I have more resources (hardware wise) to utilise.
Yes, package optimisation is a valid remark, I am checking on that end also. But is there any way to introduce more parallelism? Could there be some reason that because a package is being invoked (not a function), the parallelism isn't working?
October 25, 2017 - 7:54 am UTC
Could there be some reason that because a package is being invoked (not a function), the parallelism isn't working?
As I said before:
"That *very much* depends on what the *content* of your PL/SQL function contains and whether it is a suitable candidate for parallelising"
You can simply "tell" the database a plsql routine is fine for parallelism (via parallel_enable) but you need to make sure this is indeed valid (ie, check the docs about session state etc).
But I would trace the code - see *where* the time is being lost. And then act on that.
Need to use parallel_enable option
Sankalp, October 24, 2017 - 3:27 pm UTC