Migration to 8.7 + Oracle connectivity

[ponzio]

Hi.
I'm planning to migrate to Datasatge 8.7.
I need a hint to choose Oracle Client to install.

In my old datastage server I connect both to Oracle 10g and Oracle 11g using
Oracle Client 9i (32 bit).
This is my production environment and this scenario works well except for a little problem with numeri character described here:

viewtopic.php?t=142796&highlight=

My new server is Windows 2008 R2 Standa 64 bit.

Which Oracle client should I install ?

Thanks,
Andrea

[rprajput]

See the special configuration advice for Windows 2008 and 64 Bit systems - The DataStage job stages are 32 bit on both Windows 32 bit and Windows 64 bit installations. Consequently, all client libraries needed for connectivity (for example Oracle client support) must be 32 bit.

Hope this helps!

[ponzio]

Thanks !

I'm using Oracle Client 9 ... do you think I should use 10 or 11 ?

Andrea

[ponzio]

I tryed to install Oracle 9i on Windows 2008 but it prevented me to end this task.

I've found this that seems to confirm rprajput (even if this article is not related to OCI).

I try to install Oracle Client 32bit (win32_11gR2_client).

Andrea

[PaulVL]

We're upgrading our shop to Oracle Client 11ish. Found a bug in 10g. Mind you my client is on Linux database server is AIX.

[ponzio]

I've Installed Oracle Client 11g.
I'm using OCI9 to connect to DBs and I'm connecting to Oracle 10g and 11g DBs.

I have some problems loading an unicode Oracle 11g using OCI:
- italian data are loaded fine
- bulgarian data are not loaded properly

If use "Oracle Connector" instead of OCI the bulgarian dta are loaded fine.

IBM support is helpig me ... I let you know.
Andrea

[ray.wurlod]

Help us to help you. What's different about Bulgarian data - accented characters perhaps?

[PaulVL]

I'm guessing this is the same issue as your NLS thread...

if so, please close one and only have one open to help drive it to root cause.

[ponzio]

Help us to help you. What's different about Bulgarian data - accented characters perhaps? ...

Bulgarian is Cyrillic (see http://en.wikipedia.org/wiki/Cyrillic_script) quite different from italian ...

I'm guessing this is the same issue as your NLS thread...

if so, please close one and only have one open to help drive it to root cause.

You're right ... To be precise the solution could be the same but I've understood this at the end.

The problem with Oracle OCI is that the Client Oracle itself was installed with
ITALIAN_ITALY.WE8MSWIN1252 (as you can see in Windows Registry)
and when I try to load data in a different language it try to do the conversion to this 1252 anyway ... even if I use the right NLS 1251 in my OCI stages.

Setting up the environment variable NLS_LANG=BULGARIAN_BULGARIA.CL8MSWIN1251, that for my purposes I've created at project level (in Datastage Administrator Client), solved the problem.

You said the problem could be the same with my DB2 (see viewtopic.php?p=423298#423298)
IBM DB2 support told me that the problem could be a difference between server and client fix version ...
I'm going to check this and let you know!

Following the same logic of Oracle Client,
creating the environment varibale DB2CODEPAGE=1251 makes my db2 extract query run ok .. so this is the same solution.

Andrea

[ponzio]

Now my ultimate problem is Windows System Locale of Datasatage Server.

I'm going on my other thread viewtopic.php?p=423390#423390
with widows system locale issue and to show my checks with db2 fix test.

Many thanks,
Andrea