IBM Mainframe Forum

I have to backup a list of data sets, probably about 200 on each system we have. It there a way to have JCL read in a list from a flat file then go out and back them up? I'm really new to JCL so I'm still working out a lot of its abilities.

Unshriven wrote:I have to backup a list of data sets, probably about 200 on each system we have. It there a way to have JCL read in a list from a flat file then go out and back them up? I'm really new to JCL so I'm still working out a lot of its abilities.

Note that JCL will not and cannot do this; however, a program executed through the medium of JCL can.

Now, off the top of my head, there is no utility that will take a mere list of DSNs and back them up (to what?). A deck of control cards is needed, and probably a little Rexx, with or without ISPF, or DFSORT/Syncsort to take your list and substitute the names for the appropriate tokens.

There are many utilities that can be used for backing up data sets. Do you have a preference or a special requirement as to which is used?

Hello,

Suggest you talk with your storage management people about using ADRDSSU to backup/restore these files.

Keep in mind that once created, a backup must be used (at least once) to confirm that all that is needed has been successfully copied to the backup media.

ADRDSSU supports wild-cards and renaming of datasets (renaming can be handy for verification testing).

Questions that have not yet been answered (or asked, for that matter)

• Backed up to where? Disk or tape?
• If the backup is to tape, do you want the data to "hide," as it were, in a container, or do you want the backup datasets to be unique datasets themselves?
• Are the datasets currently on disk or tape?
• Do you want the original datasets to disappear, but in a way that they can be automatically restored if they are needed?
• How long should the backup be retained?

These questions must be answered before the backup method and utility can be selected.

Dick Scherrer proposed the use of a particular utility that is present in most shops. ADRDSSU is a fine program with excellent documentation, but it cannot handle all of the bullets in my list.

One detail that is often forgotten is make 100% certain you can recover the datasets from backup. This detail is often forgotten; if it cannot be done the most sophisticated backup method is useless.

Hello,

And an unused backup is merely some media with questionable content. I have had several clients that asked me to look into why their restored data (restored because of some other crisis) was not correct/usable.

IMHO - 2 problems here
1. before restoring, a backup of the current data (all of that is to be restored) should be backed up first.
2. the initial backup was untested and was incomplete/incorrect which would have been caught before there was a crisis . . .

While ADRDSSU might not do everything for every system, it will do more that 90% (more like 99% if done well in combination with SMS) in more than 90% of the IBM mainframe environments.

( IMHO again ) the backup/restore can be elegant, but should not be complicated. Most times it will be used "for real" there will already be some processing disruption and "getting things back together" needs to be most smoothe.

Agreed. The simpler, the better, especially when you have to reassemble Humpty.

For our fairy tale challenged colleagues: there is a English language fairy tale about an egg (Humpty) having a fall, and it proved to be difficult, perhaps impossible, to put Humpty back together again. This is a major problem if the data is being altered during the backup.