The archive file will be closed when the last record is written into it, no matter how much records or file size you have specified.
SARI does not necessarily access the archive file itself. you create an index on the archive and this index is nothing else than a table in your SAP system, which holds the most important fields of the archived object ,e.g. the key fields and a link to the record in the archive.
SAP has already given some data structures for this index, it is on you if you use it or build an own index. The more fields you add to the index the bigger will be this table and the less positive effect to the overall space usage has the archiving itself. Without index you have no quick access as the entire archive file needs to be read sequentially to retrieve the selected data. You have to evaluate the pros and cons to find the right solution.
In our system analysis are usually made from a BI system, only a few users have access to the archived data, we usually hold 2 years online, older data gets archived. The demand to access archived data is so low that we do not have indices for most of the archiving objects. And in those exceptions when people have to check old documents they accept a longer runtime to see the records.