At first, I think you're confused, the number of rows will increase linearly, not exponentially, which is a big difference.
The second 100k is nothing for the database. even if you store 365 days, which is only 36 million, I have it in a week,
A third store in JSON may complicate a future request.
Therefore, I suggest going with the approach 1
Using its own index, design, and fast hard drive, db can handle a billion records.
You can also save historical data in a different scheme, so the current data is a little faster, but this is just a minor tuneup
source share