Next
Previous
Contents
Here is a brief list of the steps needed to carry out the migration
from MVS to Unix. Do not be deceived by the apparent order and logic
of the steps in the list. Unexpected stumbling blocks were found
at each step that delayed the completion of the step or required a
change of tactics. If you are carrying out a migration yourself,
expect the unexpected.
- write a bare-bones JCL to Perl translator (expect to be only running
SAS programs)
- prepare some JCL scripts (by removing special formatting)
- transfer (FTP) some test scripts from MVS to Unix
- debug the text file preparation and transfer process
- translate some test scripts
- debug the translator until the test scripts run
- debug test scripts to remove system incompatibilities
- prepare enough support files (
CPORT
ing of SAS datasets) to get the
test scripts producing correct results
- transfer support files from MVS to Unix with ftp
CIMPORT
the support files into development directories
- run the scripts with the support files
- debug the support file preparation and transfer process
- translate (by hand) the STAT menu system to a Unix shell script
- debug the translated menu system until it runs
- prepare to erase or change
TSO
calls in SAS programs
- prepare all JCL scripts for transfer (get them out of cartridge
storage and remove formatting)
- prepare all support files (
CPORT
the SAS datasets) for transfer
PKZIP
all of the STAT directories
- transfer
PKZIP
ped files to Unix with FTP
unzip(1)
files into the /data/stat/test
directory
- (OPTIONAL) create symbolic links in
/
to emulate MVS prefix structure
CIMPORT
the SAS datasets
- translate accented characters in NOTES file (and any other files) via
SAS export,
tr(1)
, and then SAS import (if this is not
acceptable CPORT
will have to be fixed by SAS Institute or
TRANSLATE option of CPORT
must be used)
- run JCL to Perl translator on all scripts
- erase or change TSO calls in SAS programs
- test and debug currently-used scripts (leave the others for STAT to
check and debug at a future date)
- finish debugging the STAT menu system
- resolve printing solutions (maybe a big job, probably not)
- test and debug the ensemble with V and an "ordinary" STAT user
- arrange daily backups with system administrator
- copy everything to
/data/stat/prod
and check
- stop all data entry on MVS
- prepare all of the latest support files (
CPORT
the SAS datasets) for transfer
PKZIP
all of the latest STAT directories
- transfer latest
PKZIP
ped files to Unix with FTP
- unzip latest files into the
/data/stat/test
directory
CIMPORT
the latest SAS datasets
- translate accented characters in latest
NOTES
file
- test to make sure everything is still working correctly (use a
checklist of previous bugs as a regression test)
- copy everything to
/data/stat/prod
and check
- create accounts for data entry personnel
- do a demonstration for the data entry personnel and give each of them
a copy of this report
- data entry starts again on the Unix system
- *** END OF MIGRATION ***
- Next project: replace DCF report generation with LaTeX
Next
Previous
Contents