Re: SQL CLI and performance

On 6/20/05, Nathan Andelin <nandelin@xxxxxxxxx> wrote:
> 
> Under the SQL CLI interface, you can do file
> overrides, just before generating a result set, but
> I'm not sure how it might improve performance.


I was thinking about using SEQONLY(*YES XXXX) on the override to improve 
block sizing of the file accessed by the SELECT statement. When reading all 
or a great number of records, this works well for RPG, not sure about SQL. 

I've noticed quite a bit of "initialization" overhead
> associated with SQL CLI, but runtime performance is
> comparable to what I get while using SQL
> interactively, via the STRSQL command, which has been
> "good enough" for me. Are you comparing
> initialization, or runtime performance?


I am comparing total runtime, so all initialization tasks are included. In 
the test I did, I ran the program a few times before recording the time. I 
wanted to how the app would run in production. 

I've never run into a case where traditional I/O
> couldn't beat an SQL statement.


I have always heard that SQL was faster also. It must be the initialization 
phase that is slowing it down. Thanks for the info.