Hello all,
We are having a DB2 V7.1.2 on IBM Z/OS.[will be migrating to DB2 V8 soon].

And DB2 on the Windows platform :

Product identifier SQL08010
Level identifier 01010106
Level DB2 v8.1.0.36
Build level s021023
PTF
Java development kit (JDK):
Level IBM Corporation 1.3.1

Since we are in an Arab country (kuwait), I am trying to load
some arabic character data to our mainframe tables from the windows
platform. I loaded some data from SQL SERVER to my DB2 table on
windows; the Arabic is fine when the data is in my windows DB2;
when I try to export the same data from my DB2 table on windows
to an IXF file and then import to my table on Z/OS,
I find the Arabic character data is NOT SHAPED (ie, you might be
aware that the Arabic alphabets will take different shapes depending
on its position in a word) when I check that data on my PC3270
Emulator screen (which again is IBM Personal Communications v5.9
for Windows).

The command I used to import is :

db2 => import from wpperson1.ixf of ixf messages impozos
insert into kazik.tperm_sam

Number of rows read = 98
Number of rows skipped = 0
Number of rows inserted = 98
Number of rows updated = 0
Number of rows rejected = 0
Number of rows committed = 98

I can see the hex values for the Arabic characters are different on the two platforms.
According to the IBM literature the code page conversion should occur;
and it occurs, but not the shaping part.


I read in DBFORUMS Discussion groups that one guy had data transfer
between Windows and Z/OS normally with DB2 7.1 on both platforms and
that he was facing problems when he moved to v8 on Windows platform.

My Z/OS CCSID = 420 [for Arabic]

I have this problem when I try to move data from and to UNICODE tables
on both platforms, as well as when I move data from my CODEPAGE 1256
Windows tables to CODEPAGE 420 Z/OS mainframe tables.

If I transfer the Arabic from Z/OS to Windows, the characters are
perfectly fine.

My question, is, am I missing something?
Or, IBM DB2 has this problem which has to be solved thru some
other ways?

Can you guide me to some references where I can get related info?

Additional info :
the hex values on my windows table :

C7CFC7D1C920E3D1C7DFD220C7E1CECFE3C920202020202020 202020202020202020202020202020
20202020202020202020 ادارة مراكز الخدمة
71

and the hex values on my Z/os table as displayed on my PC 3270 emulator screen:

D8A7D8AFD8A7D8B1D8A920D985D8B1D8A7D983D8B220D8A7D9 84D8AED8AFD985D8A920202020202020202020202020202020 71

Note : I have eliminated the Arabic part because when I copy paste from z/Os screen to my Notepad text file, the Arabic is properly presented.
But you can see the hex values are different on the two systems.

Can someone offer some direction or pointers???
thanks for any effort....

Hi,
connecting to DB2/zOS from DB2/Windows you need DB2 Connect product. What is your DB2 Connect product edition (personal or enterprise) and what is the version of Connect?

Code page translation is in response of Connect product. So first you have to know where is this product installed and which edition/version is installed.

So when you migrated from db2 v7 Windows -> db2 v8 Windows, did you also migrate Connect product?
Hope this helps,
Grofaty

Hi Grofaty,

On Windows, I have DB2 UDB v8.1 Enterprise Server Edition. I don't have DB2 Connect installed.
My understanding was that UDB would do the conversions while moving data between platforms.
Is the DB2 Connect to be installed on the same machine as the DB2 UDB is installed? Or should it be on another box and then use it to connect the databases on the Z/OS and the Windows?
I will do some search in that direction.
Anyhow, thanks for yr response.

SAMMOI

if db2 ee is installed, connect is included and not needed as seperate product anymore
the db does do the conversion, but take care of command window/pc3270 translate. the data can be stored correctly but the environment does a wrong display

As guyprzytula wrote, if you are using DB2 Enterprise Server Edition then Connect product is already installed.
Before calling IBM support I would download and install the latest fixpacks for DB2/Windows to make sure this problem is already solved.

Hi all,

As suggested by Grofaty, I downloaded and installed the Fixpack 15 for my
DB2 UDB Enterprise, and the current level is as follows :
C:\Documents and Settings\db2admin>db2level
DB21085I Instance "DB2" uses "32" bits and DB2 code release "SQL08028" with level identifier "03090106".
Informational tokens are "DB2 v8.1.15.254", "s070720", "WR21383", and FixPak
"15".

After this, I followed the recommendations for BIDI support as given in IBM manuals like Administration Guide:Planning v8 pages 258 onwards for enabling the Bidirectional support. I tried different CCSIDs for Arabic to override the Z/OS CCSID, and now I have improvement in that most of the Arabic characters come OK on the Z/OS, but few characters that come at the end of words do not have the proper shape. [It is very difficult to explain the shaping problem to someone who is not familiar with Arabic].

My question is - is it time IBM must be contacted?
We had a similar experience with Websphere MQ when IBM had to finally give some fixes on the Windows platform so that the. What is holding me back is, the documentations do not say of any similar problem at all. Is there anyone who is successfully transferring Arabic characters between DB2 on Z/OS and windows platforms?
Any suggestions or pointers to source of relavant technical info will be greatly appreciated.

SAMMOI