Re: PF Compiled Files with Dictionaries vs. SQL-Created Tables

My claim about not seeing SQL perform better than RLA also applies to SQL
stored-procedures that are written in a pure SQL language. We've had to
rewrite a few of those stored-procedures to now use external RPG program
(without creating any extra logical file) to get an acceptable performance.


"Jason Abreu" <jason.abreu@xxxxxxxxxxxxxxxxxxxx> wrote in message
news:mailman.26775.1296660342.2702.midrange-l@xxxxxxxxxxxx...

But embedding SQL in a HLL is a different animal than executing a SQL
statement vs a HLL program.
This is evidenced by the need for a precompiler to translate the embedded
SQL into calls the HLL can understand.

RLA is faster than executing a CALL to retrieve a single record when
looping through a few thousand records. And that is, in a rough sense,
what embedded SQL is translated into.

Jason Abreu
Abreu Innovations, Inc.
jason.abreu@xxxxxxxxxxxxxxxxxxxx
http://www.abreuinnovations.com/

On 2/2/2011 10:18 AM, hockchai Lim wrote:

didn't read the article. But I do know that IBM and some folks have
claimed
that SQL is faster than RLA. But from my real live experience, it is
hardly
the case. I use embeded SQL in RPG applications quite frequently.
Almost
in all cases, the response time in those applications are either about
the
same or worse than the RLA version. May be I didn' t create the SQL
statement correctly, may be I need to create some indexes to make it run
faster, or may be I need to fetch in block... so many possibilities. Now
who is got the time to do those guess work :). Don't get me wrong, I
love
using embeded SQL in RPG. But I just don't see it perform better than
RLA.


"Gary Thompson"<GThompson@xxxxxxxxxxx> wrote in message
news:mailman.26728.1296655165.2702.midrange-l@xxxxxxxxxxxx...

You might want to know about relative performance of tables created
using
SQL DDL vs DDS:
http://www-03.ibm.com/systems/resources/systems_i_software_db2_pdf_Performance_DDS_SQL.pdf





CRPence<CRPbottle@xxxxxxxxx>
Sent by: midrange-l-bounces@xxxxxxxxxxxx
02/01/2011 06:05 PM
Please respond to
Midrange Systems Technical Discussion<midrange-l@xxxxxxxxxxxx>


To
midrange-l@xxxxxxxxxxxx
cc

Subject
Re: PF Compiled Files with Dictionaries vs. SQL-Created Tables






On 2/1/11 1:13 PM, hockchai Lim wrote:

I know I'll probably get blasted for saying this. But I must admit
that I've never been a big fan of using SQL to create tables. I
understand that using SQL will allow me to create table with feature
that is not supported by DDS (such as Identity column). But that is
IBM's fault not mine :). The main reason I didn't like SQL is because
it strains my eye to read that SQL create table statement. DDS is
just much neater and easier to read.

And DDS even supports some features which SQL does not. I find
little value to change from DDS to use SQL DDL for existing database
*FILE objects, without some obvious need to access some new feature
provided by the TABLE; e.g. either data types supported only in the SQL,
or better integrity for numeric data. Without any specific goal to be
achieved by changing, what point is there in change except change for
the sake of change? That is, change without justification is
intuitively, not justified. Best to understand what will be lost and
what will be gained by the change, to avoid remorse for having made the
jump, and even more regrets for having to move back in order to
"recover" from the first change. Changing "just because" to using SQL
DDL versus DDS is, for lack of a better word, daft.

While many DDS PF source may be very clean and concise, I mostly
disagree that DDS specifications would be generally easier to read than
the SQL, except to those who have some experience coding to fixed-format
source specifications. I think difficulty in reading the SQL DDL would
probably be caused mostly by poor formatting of the statements,
exacerbated by [typically self imposed] limitations to line length.
FWiW the LABEL ON statement for column headings and text being moved
outside the functional field definitional attributes could be considered
by many to be much nicer than having them intermingled and even
line-wrapped from and back into the functions area of a DDS source
member.

Regards, Chuck
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.