DataStage MVS Edition

gstungare · Post by **gstungare** » Tue Jul 17, 2007 9:46 am

Hi,

We have source and target as MVS DB2 tables. Datastage server is currently installed on a UNIX box

Could someonelet me know if we have to install 'IBM WebSphere DataStage MVS Edition' and give some info abt it

Gaurav

ray.wurlod · Post by **ray.wurlod** » Tue Jul 17, 2007 3:04 pm

You don't need to install MVS edition, though your IBM account rep would be happy for you to do so. You can get the data from the mainframe using FTP.

dls · Post by **dls** » Tue Jul 17, 2007 3:08 pm

If source AND target on the mainframe, then it does make sense to use the MVS product.

lstsaur · Post by **lstsaur** » Tue Jul 17, 2007 9:28 pm

Gaurav,
No, you don't need MVS version. All you need is to install the DB2 Client in the Unix box and do all the Catalogs and Bind and have proper permissions to access your mainframe DB2.
I have been using either ODBC stage or DB2 stage to access MVS DB2 database.

lstsaur · Post by **lstsaur** » Tue Jul 17, 2007 9:29 pm

Sorry, I meant DB2 Connect 32-bit client.

eostic · Post by **eostic** » Tue Jul 17, 2007 9:37 pm

This is always an interesting decision.....if a site has mainframe sources and mainframe targets, and the transformations are such that you want to do the crunching on the mainframe (some sites would just as soon NOT have any heavy processing there), and/or you have complex VSAM/IMS/DB2 sources that you'd like to source in the same job, and you have a very z Series aware shop that is comfortable with COBOL, and Unix is a four letter word at the site, or your mainframe admins don't want sensitive data floating between machines........ then MVS Edition can be a very wise choice...... but from a pure technology perspective, there are connectivity tools that will let me move data between sources/targets on the largest of mainframes with only my laptop...... lots of factors to consider.... is this just one job or set of jobs that is required, or a whole new project in an area that finds Unix/Linux repulsive (some sites do...)?

Ernie

ray.wurlod · Post by **ray.wurlod** » Tue Jul 17, 2007 10:36 pm

DataStage MVS generates nice COBOL and JCL, but the politics of getting "them" to allow it to run on their precious mainframe can be altogether too time-consuming.

gstungare · Post by **gstungare** » Wed Jul 18, 2007 7:43 am

eostic wrote:This is always an interesting decision.....if a site has mainframe sources and mainframe targets, and the transformations are such that you want to do the crunching on the mainframe (some sites would just as soon NOT have any heavy processing there), and/or you have complex VSAM/IMS/DB2 sources that you'd like to source in the same job, and you have a very z Series aware shop that is comfortable with COBOL, and Unix is a four letter word at the site, or your mainframe admins don't want sensitive data floating between machines........ then MVS Edition can be a very wise choice...... but from a pure technology perspective, there are connectivity tools that will let me move data between sources/targets on the largest of mainframes with only my laptop...... lots of factors to consider.... is this just one job or set of jobs that is required, or a whole new project in an area that finds Unix/Linux repulsive (some sites do...)?

Ernie

Thanks for the Response

There are about 100 DB2 tables on mainframes source sytem . we want to move to target DB2 tables which is also on mainframe

gstungare · Post by **gstungare** » Wed Jul 18, 2007 7:54 am

lstsaur wrote:Gaurav,
No, you don't need MVS version. All you need is to install the DB2 Client in the Unix box and do all the Catalogs and Bind and have proper permissions to access your mainframe DB2.
I have been using either ODBC stage or DB2 stage to access MVS DB2 database.

Hi Thanks for the response

So, although I want to move data from one mainframe db2 tables to another m/f db2 tables , I donot need MVS Stage
But I need to ensure that db2 client on unix box have proper access to all m/f db2 tables

Gaurav

dls · Post by **dls** » Wed Jul 18, 2007 8:03 am

You can find plenty of cases where clients are sourcing from DB2 on the mainframe, but staging, processing, and loading targets on Unix platform.

If anyone is sourcing and targeting DB2 on the MF, but processing on Unix or Windows, I haven't seen or heard of it. If there are, and unless it were a simple data migration effort, I'd be interested to know their rationale.

Last I heard, running MVS edition with USS, had some areas of unproven reliability. That may have been remedied by now.

Sound like you're an ideal candidate for a proof-of-concept.

eostic · Post by **eostic** » Wed Jul 18, 2007 10:43 am

...this is why it's such a complicated and many times, "cultural" issue. As Ray states above, politics may prevent it ---- but equally, politics may prevent the alternative. This is especially true for sites where there is no Unix (or very little), a team that is comfortable with COBOL (there is no need to review the generated COBOL, but it's comforting for such sites to know that all the I/O, methods, security paradigm, etc. are doing "normal" COBOL activities). If a site ALREADY has DS experienced on LInux/Unix/Win and lots of skills there, and has one project to do via mainframe, the decision to use MVS Edition is less likely. There are sites that use both, but often it's because they have departmental separation where each department still fits a particular culture. There is no right or wrong answer here, only the right one for the performance, political, re-use, cost, etc. needs....

Ernie