I am working on a project to regularly process terabytes of legacy System z (mainframe) data (e.g. VSAM, Physical Sequential)) with DataStage. DataStage will be co-located on System z running in z/Linux. Communications between z/OS and z/Linux will be via HiperSockets.
https://www.ibm.com/docs/en/zos-basic-s ... persockets
For Physical Sequential (PS) datasets we plan to use the Linux zdsfs command to mount z/OS DASD for direct read by the Sequential File Stage.
https://www.ibm.com/docs/en/linux-on-sy ... ands-zdsfs
https://www.ibm.com/docs/en/iis/11.7?to ... sequential
The zdsfs approach comes with some limitations:
- an individual file can't span a z/OS volume
z/OS catalog services and auditing (security) mechanisms are bypassed
https://www.ibm.com/docs/en/iis/11.7?to ... mplex-flat
It appears from the documentation that the VSAM can't be natively read with CFF from DataStage. Even with HiperSockets, copying terabytes of data via FTP from z/OS to z/Linux creates a lot of data movement and will require vast amounts of z/Linux persistent storage.
I am writing to perform a sanity check that this is the best approach this challenge.
TIA,
Frank