Mainframe, MVS and zOS Discussion

Could you please explain the differences and advantages of using DSN=DUMMY versus DSN=FILENAME, where FILENAME is an empty sequential dataset and an empty VSAM dataset?Additionally, is there any difference between using DD DUMMY and DD DSN=NULLFILE?This was an interview question, and despite referring to some JCL books, I couldn’t find a clear answer.

Hello.

First of all, the dummy syntax is "DD DUMMY" not "DD DSN=DUMMY". Coding "DD DSN=NULLFILE" is essentially the same thing as coding "DD DUMMY", it just takes slightly longer to type!

When "DD DUMMY" is used on an input DDName, it simulates an empty file. So, if you needed to, say, create an empty member in a PDS, or clear the data from an existing dataset, you could just run IEBGENER with SYSUT1 set to "DD DUMMY" and SYSUT2 set to the output and it would perform those tasks.

The difference between "DD DUMMY" and "DD DSN=empty.dataset" on an input is simply that you need to have an empty dataset allocated which is consuming disk space. Attempting to read either will result in the same outcome; end of file.

As an output, "DD DUMMY" becomes a null file. It is the equivalent of the "NUL" file under Windoze or "/dev/null" in Unix/Posix systems. You can write as much as you like to that particular DDName and the records just disappear. You'd use this in the scenario where a program is writing data, but you don't need all or part of the output from it.

"DD DUMMY" can also be used with VSAM files. Personally, I've not seen this done and have never had to use it, so I can't really comment any further on that scenario.

For answers to JCL questions, the best first place to look are IBM's manuals:
- MVS JCL Reference: https://www.ibm.com/docs/en/zos/3.1.0?t ... -reference
- MVS JCL User's Guide: https://www.ibm.com/docs/en/zos/3.1.0?t ... sers-guide
 

The difference between "DD DUMMY" and "DD DSN=empty.dataset" on an input is simply that you need to have an empty dataset allocated which is consuming disk space. Attempting to read either will result in the same outcome; end of file
 

NOPE

when processing a single dataset that is , by pure chance, true
but when processing a dataset concatenation the behavior is different 
processing will stop at dsn=nullfile and the third dataset will be ignored

but having
the process will  process ENRICO.TEST.ZB.EMPTY  returning a normal eof
and continue with ENRICO.TEST.ZC

just tested ....
 

First of all, the dummy syntax is "DD DUMMY" not "DD DSN=DUMMY". Coding "DD DSN=NULLFILE" is essentially the same thing as coding "DD DUMMY", it just takes slightly longer to type!

Thank you for correcting it, understood.

As an output, "DD DUMMY" becomes a null file. It is the equivalent of the "NUL" file under Windoze or "/dev/null" in Unix/Posix systems. You can write as much as you like to that particular DDName and the records just disappear. You'd use this in the scenario where a program is writing data, but you don't need all or part of the output from it. 

But if I use an actual empty file, the data will still be written, now as coder I should decide to keep it or not?

Yes, if you use an empty file and the program you are running writes data to it, the data will get written to the once empty dataset.