This is a guide for deciding what parameters need to be coded on a DD statement. Sometimes we need many parameters, and sometimes we need just a few.

If the data set is new
   If it is a permanent file
      Use DSN, UNIT=SYSALLDA, SPACE and DISP=(NEW,___,___)
   Else  (*  It is a temporary file.  *)
      Case
         The data set is to be printed:
            Use DD SYSOUT=*
         The data set is in-stream data:
            If the data contains lines beginning with // or /*
               Use DD DATA, DLM=__ and use __ as the delimiter
            Else  (*  The usual delimiter will do.  *)
               Use DD * and use /* as the delimiter
            EndIf
         The data set is to be on disk:
            If the data set is needed in a later job step
               Use DSN=&&___, UNIT=SYSALLDA, SPACE and DISP=(NEW,PASS)
            Else  (*  We can throw it away after this step.  *)
               Use UNIT=SYSALLDA and SPACE
               (*  There is no need for DSN or DISP.  *)
            EndIf
      EndCase
   EndIf
Else  (*  It is old.  *)
   If it is a permanent file
      If it is catalogued
         Use DSN and DISP=(SHR,___,___) or DISP=(OLD,___,___)
      Else  (*  It is not catalogued.  *)
         Use DSN, UNIT=SYSALLDA and DISP=(SHR,___,___) or
           DISP=(OLD,___,___)
      EndIf
   Else  (*  It is temporary, inherited from an earlier step.  *)
      Use DSN=&&___ (name from earlier step), UNIT=SYSALLDA and
        DISP=(OLD,___)
   EndIf
EndIf

This applies to ordinary sequential data sets, not to VSAM clusters.

If we are dealing with VSAM clusters, we do not normally use DD statements to create them or delete them (although this can be done with additional DD parameters). Instead we use the IDCAMS utility. All VSAM clusters created with IDCAMS are catalogued automatically.