Allow Omegamon AI for Db2 to rollout maintenance or upgrade by LPAR. Currently it's PLEX-wide, impacting customer rollout.

Overview
The current design of the ISPF interface/FPEJINIT code prevents lpar-independent upgrades.  IBM has for decades promoted lpar-independent upgrades.  An opportunity exists to bring Omegamon for Db2 into the same migration family.  

ISPF calls a lpar-specific panel, in this case DB2<rte_name>, which calls FPEJINIT who allocates the runtime libraries in hlq3.1 and hlq3.2.  These are version specific, regardless of hardcoding or alias.  

We prefer to roll out code by lpar, allowing us to provide our customers with zero downtime.  We have not had an outage in 15-20 years.  It's been so long we lost track. 

With a shared Db2 catalog for all members in the data sharing group, the bind is shared across the lpars.  Different versions can exist concurrently in the catalog.  

The issue is FPEJINIT.  With a single hlq3.1 and hlq3.2, it is not possible to rollout Omegamon AI for Db2 by lpar as all lpars are affected simultaneously. 

Once FPEJINIT is updated (single hardcoded name called by the ISPF panel), it will allocate the runtime libraries.  All lpars call the same member.  

As you can see, the FPEJINIT single piece of code allocating the runtime library forces a plex-wide upgrade.  

For simplicity and stability, manual updates of the architecture are discouraged.    

I can create versions of FPEJINIT to allow LPAR upgrades, however this violates our policy of installing as shipped.    

Method to resolve
The ISPF panel contains a variable component for the lpar name; the same logic can be employed for FPEJINIT.  
FPEJ<rte_name> would be the new variable, and a corresponding update to the ISPF panel calling the lpar-specific FPEJ<rte_name>. 

Business Justification
Customers are forced to a non-conforming methods and variables, possibly introducing unwelcome complexity for upgrades and maintenance.  They are also forced into a plex-wide upgrade due to the missing FPEJ<rte_name> variable, and potential outages when the wrong released code is called in FPEJINIT. 

Introducing the simple variable as described above will eliminate the plex-wide upgrade, allowing customers to rollout maintenance lpar-by-lpar, following the convention of the rest of the IBM product families.