ODBC for Cobol?

[#1]

Hi,

I 'm need a connection ODBC of Cobol whit QlikView, if anyone knows would appreciate your help.

Thanks

Att: Diego

[rwunderlich]

There is no COBOL ODBC exactly. Typical COBOL data access would be a DBMS like DB2, or VSAM or flat files. I'm guessing you are trying to access VSAM or flat files? Can you be specific?

-Rob

[#3]

Thank's for your help.

I' m basic need is extract the data .dat and .inx of Cobol and convert on .qvd in QlikView, or of know a tools that migrate the data of Cobol to SqlServer or Oracle.

Thank's

Att: Diego Hernandez

[hector]

Hi

For your need exists the product Relativity, it converts the Cobol Files into "tables", that can be accesed by an ODBC connection.

I've participated in projects with this tool with good results

Rgds.

[#5]

Hello !

We use Cobol on mainframes here. And we use QV with all of our mainframe files.

What we do is to extract the tables as CSV files. We "dump" mainframe tables as CSV files outside the mainframe.

And we only "dump" tables that have been modified since last dump. This optimizes a lot the download process.

We also have built a kind of "automator" that reads the table layout and builds automatically a Cobol program to dump that table into a CSV. The developer simply points the tables that are to be downloaded and also the fields to be downloaded. All the rest is done automatically.

Anyway, let me know if I can help you with some ideas.

See ya !

[rwunderlich]

diegofha wrote:I' m basic need is extract the data .dat and .inx of Cobol and convert on .qvd in QlikView, or of know a tools that migrate the data of Cobol to SqlServer or Oracle

I'm familiar with mainframe COBOL and this looks like Windows, so I don't have any insights. However, it looks like you got another possible response.

-Rob

[#7]

Hello Diego !

Some points for you :

• I think you only need to deal with .DAT files. .INX are INDEXES files. Indexes don't need to be transfered to QV, AFAIK.
• What we done here (I'm repeating) is to have a kind of a GENERIC program that dumps a Cobol table to a CSV file. Then, we have also a GENERIC script that converts that generated CSV file into a QVD. Our QlikView final scripts, the real ones, read these latest QVD files. QVD are very, very fast. I think this approach is BETTER than migrating to a SQL server. It's simpler and faster . Simpler to implement and faster to work with daily. IOW we have a repository (a database) of QVD files. For each Cobol table we have a folder on the file server where we store the CSV file, the script that converts it to QVD and also , the QVD file itself.
• In our case all the process above is done automatically. We've built a Mainframe Data Extractor Tool. But you can do it pretty well manually. Do it once, study the program etc and then, try to transform it to a INCLUDE or PROCEDURE or SUBPROGRAM (CALL). Or simply cut and paste it (replicate it) for the other tables.
• As for the CSV files, we keep them on the file server because they are easy for you to open up and see the data. But they could be discarded once they are converted to QVD. I prefer to keep them. You can also ZIP them.
• We like CSV nicely formatted that is, "easy" to see. Human readable (easily).
• I have attached a QVW sample that reads a CSV as we like them here and convert it to QVD. This is a tiny application showing you the concept only. This is inside TEST169.ZIP.
• Have also attached a REAL kit for you. It's a real mainframe file dumped into a CSV. So inside the REAL.ZIP you have the mainframe Cobol program that reads the original table, builds the CSV file for it, transfer it outside the mainframe via FTP , then invokes the QV script to convert it to QVD and STOPs RUN. Also inside the kit the CSV file and the QVD file.

Hope that all can help you, OK ?

Where are U talking from ? Maybe we can exchange ideas in private mode (email).

Good Luck ! See ya !

[#8]

Forgot to mention : our source files are in OEM coding !

[#9]

Hi, Adriano

Thank's for your help, I Think will be very useful.

See ya !

[#10]

OK.

One ESSENTIAL thing that you must understand : ALWAYS transform your CSV files into QVDs.

QVDs are way, way faster to be read by QV.

Since your final applications will have to read the files several times during development process, perfomance here is very important.

Have done some tests here. You can have an idea :

• Reading a 500 lines file from CSV : 4 minutes and 25 seconds
• Reading THE SAME FILE above converted to QVD : 18 seconds !!!

So, hope you understand this key point, OK ?

Let us know what you ended up doing there on your side, OK ? So that we can share some ideas as well.

What environment you are ? Mainframe ? IBM ? What is the machine and Operating System ? We're on a Unisys Mainframe.

See ya !