I was wondering if there are any best practices for migrating data from Mainframe to Unix, considering also a version change from 7.1.3 to the current version for Unix.
Are there any approx. times for migration? (i.e. hrs/GB or something like that?)
We have a very limited window to perform the migration.
Thanks in advance.
For instance, do you know any product that works with adabas that allows hot replication of data? We have a “custom” PLOG behavior, and we are using 3 files instead of just one, in order to have more history on PLOG data.
Another “issue” we have is that the Adabas fields do not exactly match the underlying data. We have fields of A(10) in the DFT, for instance, that are redefined in the programs as A(5) and N(5), which requires a specific conversion algorithm. Regarding this “problem” I was wondering if it was possible to actually create new fields in the DFT on Unix to match the redefinition used by the natural programs. Is that possible? will the adabas load tool migrate the data correctly? Will programs still compile / run if the redefinition of fields used inside them now actually have a corresponding physical file in Adabas? (We would also be creating a field hierarchy in the DDM in order to keep the same names)
Let me know if this is not clear and maybe I can clarify with an example.
Thanks in advance.
“Event Replicator for Adabas” does this.
do you know it will handle the multiple PLOG file stuff as well?
On your redefinition issue - yes, it is possible to convert the redefined fields into two and have them load correctly. To migrate the data, you have to unload and decompress the data, transmit it to your other platform, compress and load there. So, during the compress step, you can introduce your new fields.
Lots of caveats of course:
- if you are loading to a numeric field, the source data must be numeric. This is not guaranteed when you are unloading from an alphanumeric field.
- you will need to change your programs to take advantage of the new fields. For backward compatibility you could also introduce a superfield with the name/length of the original field.
I’m not clear on what your PLOG question is. What are you currently using the PLOG data for (other than recovery)? Event Replicator does not use the PLOG - it does near real time replication of data, replicating at ET time.
Event replicator for Adabas can do replay from PLOG. But it is also unclear to me what you want to achieve?