The easiest way for transparency is to have a single PI System.
#1: You could use some automation (e.g. PowerShell) to form dynamic archives that each cover the time range of 1 month; 12 archives in total. Then periodically have a script that un-registers an archive that covers a time period older than 12 months.
#2: Nothing elegant. You would have to embed that logic in your application retrieving the data. If it is off the shelf OSIsoft tools then there is no easy option other than a single system.
Why do you have the two systems?
Many databases experience performance issues after the data is built up over time. This is a common reason we have seen for this kind of question. However, the PI Server is designed with an indexing system that a query of a period of data 5 years ago is just as fast as a query of that same time period starting 5 days ago. The only time we see older data queries take longer is if the older archives are relocated across a network or to a different machine.
Except in special circumstances you should be able to keep years worth of data on a server with no performance issues.
Some organizations have data retention rules that make it necessary to delete old data (after a fixed time period), but that cannot be the case here as you say you want to keep it.