Re: RPG Blocked Writes

The help text for the SEQONLY OVRDBF parameter has a lot of information about this, as I recall. It speaks of the various conditions when record-at-a-time is used. Might be worth a look.

Also, the Database Programming manual has a fairly extensive section on SEQONLY - if you don't specify NBRRCDS, it says that if there are ANY other access paths to the PF, it gets changed to SEQONLY(*NO) - no blocking at this level. They don't have to be unique.

BTW, I'm wondering if you are confusing physical I/O (storage to main memory movement) with system to internal buffer movement when discussing page sizes. Not sure - and it's been too long that I've thought about this. But I thought there's a double clutch effect, or used to be. This involves the IO from dasd to DB buffers, then from DB buffers to program buffers. Some of this has changed, right? With input to data structures?

I am soon finding the limits of my real knowledge here!! But I had to ask! :-)

HTH
Vern

On 6/7/2010 6:03 PM, Kurt Anderson wrote:

In this particular case, the physical is keyed, as are some logicals, although none of them are Unique keys.

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Dennis Lovelady
Sent: Monday, June 07, 2010 5:50 PM
To: 'RPG programming on the IBM i / System i'
Subject: RE: RPG Blocked Writes

Not keying the physical won't help, if you have unique keys elsewhere. The
result is a constraint on the file that must be checked with each write -
and that means a physical write with each logical write.

I personally feel it's best to put your unique keys on the physical, but
again that won't impact the performance you're measuring.

Dennis Lovelady
http://www.linkedin.com/in/dennislovelady
--
"Abstract Art: A product of the untalented, sold by the unprincipled to the
utterly bewildered."
-- Al Capp

Aha, so it sounds like there is a reason to not key a physical? I
thought that the "sequential" and "keyed" access to a file as mentioned
by the document was in regard to what opcodes I was using, not the
actual access path. Bummer. However, good to know. A while back I
had asked if there was any reason to not key a PF, and I don't recall
that being an issue, however I may have also missed it.

Thanks,
Kurt

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-
bounces@xxxxxxxxxxxx] On Behalf Of Alan Campin
Sent: Monday, June 07, 2010 5:30 PM
To: RPG programming on the IBM i / System i
Subject: Re: RPG Blocked Writes

Are there unique keys on the table? That will change it to single
record
I/O.

On Mon, Jun 7, 2010 at 4:22 PM, Kurt Anderson
<kurt.anderson@xxxxxxxxxxxxxx>wrote:

I have a file defined in a program as:

FFileA O A E K Disk

> From my understanding, writes to this file should be blocked.

However

when I look at the I/O for the job, the I/O count and the RRN is

always

equal. For comparison, reading a file that's blocked has a lower I/O

count

than the current RRN.

Is this a matter of me not understanding the I/O screen (when looking

at

the job as it is running), or is this file actually writing a single

record

at a time?

According to this document, I feel that these writes should be

blocked.

https://www-

912.ibm.com/s_dir/slkbase.NSF/1ac66549a21402188625680b0002037e/d6738e1c
d37e1f33862565c2007cef79?OpenDocument

"All high-level language programs (HLLs) use blocking at certain

times and

use single record I/O at other times, based on program

specifications.

Because blocking takes less system resources to perform a single I/O,

a

program that blocks performs better and uses less system resources.

The

default for the HLL uses record blocking if opening a file for output

only

(write) or input only (read)."

Thanks,
Kurt Anderson
Sr. Programmer/Analyst
CustomCall Data Systems