RE: Simple Display file problem -- RPG400-L

From: Alan Campin

<snip>
My benchmarks showed the opposite.  SQL INSERT is 10-100 times slower
than
the native WRITE.  You will see some performance improvement using
prepared
statements, but native WRITE is still faster. Please provide an example
program and your benchmarks.
</snip>

This from IBM. Articles have repeatedly been in Iseries Network about
SQL improvements. According to IBM, differences occur because in DDS I/O
system writes are not validated but every time that you read a record
data is validated. In SQL I/O system, data is validated in on write and
not validated on read. If you are writing all the time and never
reading, you would see slower but usually you write something and read
many, many, many times.


Alan, I think you might want to do some basic research.  Defining your data
with DDL is a completely different topic than accessing it via SQL.  You can
read DDL-defined files just as easily using native I/O as you can with SQL.

I understand the confusion; even SQL experts get these mixed up.  But
really, it's this sort of misinformation that can convince people to make
bad decisions.

As far as reading data in, how often do we do single read vs reading in
multiple records? Most of the time we are reading in multiple records
joined, as least, logically to other records. You are almost always
doing the equivalent of SETLL and READ or READE all the time. Again
thinking database instead of files.


All the time!  Do you actually write business applications?  I have to chain
to get an item master, then chain to get a customer record.  I need to see
if a warehouse override exists.  And even when I do a loop, it's usually a
small number of records.

c/Exec Sql
 +  Fetch Next From ListCursor For 32766 Rows Into :dsItemRecord
c/End-Exec


I have NEVER created a 32000+ array of data structures.  I can't see why you
would except in utility programs or MAYBE some queries.  Either one of those
is a good reason to use SQL.  But this DS approach has other issues:  any
time the data you are fetching doesn't exactly match the physical file
layout, you have to manually create a data structure, and make sure it
matches your SQL statement.  It's double maintenance whenever you need a new
field, and seems to cancel out some of the "ease of use" that SQL supposedly
gives you.

Remember when doing SQL with join, the I/O only occurs once and even the
moves would only occur once in the buffer. From then on, only the
changed fields would be moved


If you read into an array of data structures as you propose, the data
structure would contain both header and detail information.  You must copy
the header data multiple times.

My bottom line is one stated by other previously in SQL vs Record I/O
discussions. SQL has changed and the AS/400 has changed. A lot of the
old assumption we used to make are just not true anymore.


But you haven't proven a single thing.  That's why I despise these sort of
mental gymnastics.  Prove your point.  Write a benchmark.  Until then, it's
just smoke.

Joe