UNC-Chapel Hill MVS-to-UNIX Migration Archive
Archive Deletion in -496 Days!
(as of Tue May 13 08:19:33 2003)
This graph shows the number of times people have used the checkout utility to copy data from the archive and how many files have been checked out since we started recording this information.
Graphs last updated Tue May 13 08:19:33 2003.
FAQ
This FAQ was intended to serve as a gentle introduction to the MVS to UNIX Migration Archive and how to use it. It was written in the early part of the migration (mid-1995) and makes references to the MVS system still being around. The Convex it keeps talking about was at the time our biggest UNIX box, and it had a large tape robot tied to it that actually stored the archived data. General user documentation was created shortly after this FAQ was written.
Neither the MVS nor the Convex machines are around anymore, the tape robot was replaced even before the migration had completed, and in Fall 1999 we moved our tape data store yet again. We had no idea at the time we copied the data from MVS that that was just the first step of this data's long journey.
It is unfortunate that the project was never formally written up in a way that could be presented as a complete package for others who might find it useful. Instead, interested persons and institutions must play the role of aercheologists, picking their way through historical documents like this one and asking questions of those forgetful people who were involved with the data movement.
To that end, I welcome any inquiries you might have about how the process worked and where we stand now. We promised our users that we would keep the archive around for five years, but we never really said when the five years started. As the designated keeper of the archive, I'm declaring its beginning January 1, 1997 -- the day after we turned off data set migration. The entire archive is up for deletion on January 1, 2002!
| What is the MVS to UNIX Migration Archive? | During 1995, OIT is copying all the user data sets from their IBM mainframe to files on the Convex. All these files are being moved into one giant archive and grouped by userid. Users will be able to checkout their formerly MVS data sets from this archive. |
| Why didn't you just move my files to my userid on the Convex in the first place? | We started to do that, but not all users have userids on the Convex, and many of those who do have customized their environment such that it was too difficult to automate the process without risking user's data. Furthermore, the coordination required between OIT staff and the individual users was far too much overhead to complete the data migration in time. By putting all the data sets in one place, we could be more reasonably sure of what had and had not been moved. In addition, by having an archive from which users can checkout their data, they need not worry so much about accidentally destroying the only copy they may have of valuable data while they learn how to work in the new environment. |
| So how do I check out files from the archive? | Run the program checkout. It prompts you for the MVS userid and password, then presents a list of data sets. Just select the data sets and/or PDS members you want to check out, and checkout will copy them into files in your current directory. |
| What if I need to checkout a file more than once? | Since checkout makes a copy of the data (the archive never changes) you can checkout any given data set as many times as you need to. Beware that any files with the same names in your current directory as those you check out will be overwritten. |
| What happens when I change a data set on MVS after you've placed a previous version of that data set into the archive? | We replace the old version in the archive with the new one. That may not happen right away, though. Data sets have to "age out" -- remain unchanged for a certain period -- before they are eligible to be automatically moved. As we get closer to the end of the year, that period will become shorter. Eventually it should get down to one or two days so that any data set you change on MVS on one day will be archived on the Convex by the end of the next day. |
| What happens when I delete a data set on MVS after you've moved a copy into the archive? | Nothing. The archive copy will remain along with all the others. |
| How does checkout know whether I should be allowed access to a given set of data sets? | It requires you to enter the userid password for the userid that owns that data set. It then asks the MVS machine if that's the right password for the given userid. If it is, then it remembers that password so that you can still checkout data after the MVS machine is no longer in service. |
| What about userids and/or data sets for which there is no known password? | We can set an auxiliary password by hand for any given userid in the archive. Then you would use that password instead of the (nonexistent) userid password. |
| What if I don't run checkout for a given userid before MVS goes away? | Then shame on you. You will have to go through the same procedures you would go through now if you were to lose or forget your password. That involves some combination of proving you are who you say you are and that you are authorized to get the data, maybe a blood test, some DNA analysis, and perhaps a bribe. At the very least, it will involve some additional time on your part in tracking down the right person, etc. |
| Perhaps I should run checkout as soon as possible on any userid for which I might want to checkout files even if I don't necessarily want to get to the files now. | That's absolutely correct. You don't have to actually checkout anything, just run checkout and give it the MVS userid and password so it can authenticate the combination while MVS is still around, then quit. It will remember the userid/password combination. You can always come back later and actually checkout files if you want to. |
| How are the different types of data sets handled during the move and during checkout? | That depends. When we move a data set, we examine its DCBs and make a best guess about what sort of data it contains. Most SAS data sets for example have DCB characteristics that identify them as SAS data sets. Depending on the type of data set, we use either ftp (in binary mode) or SAS to copy the data to the Convex. Once there, we examine some files again to further refine the data type information before we put them into the archive. Finally, checkout filters the data as necessary to make it useable on the Convex. For example, most text files on MVS use EBCDIC character encoding while the Convex uses ASCII. Checkout does the ASCII conversion as well as adding new line characters and (optionally) trimming trailing blanks off the ends of lines. |
| What if checkout guesses wrong about a file and messes it up? | You can check it out again using a different set of options, and if necessary you can check it out in binary mode and perform necessary transformations yourself. |
| How long does checkout take to run? | That depends. The part you have to deal with is quick. Once you've made your selections checkout spins off a background task to handle the actual data movement so you don't have to wait at your terminal for your files to show up. On the other hand, most of the data sets in the archive are migrated (i.e., they are on tape somewhere), so how long it takes depends on how many files you are checking out and how many tapes have to be mounted to get to your files. |
| How will I know when checkout is done? | The files will show up in your directory as soon as we start copying them out of the archive, but they will be incomplete until the copy operation is over. The best way to know it's done is by running the checkup program. We had initially planned on sending e-mail to you upon completion, but checkup is easy to use and there are more pressing issues still to be addressed, so the e-mail option has been back-burnered for the time being. |
| How can I check on checkout's progress? | The program checkup gives a report of how many files have been done so far, how many are left, how much time they have taken so far, and how long it's been working on the current file. With that information you should be able to make an extremely rough guess about how long the rest will take. Keep in mind however that it will go in fits and starts as parts of the archive are copied from tape in batches. |
| How can I find out what data sets are in the archive? | Another utility, checklist, is provided just for this purpose. It produces a listing of all the data sets and PDS members for one or more userids. Additional information about each data set (DCB characteristics, creation date, type information, etc.) is also available. Run checklist | less for more information on how to use it. |
