I have never worked in a brewery, but I have done QA in other production environments. Yes, the storage requirements can grow quickly and steadily if you plan to retain a six pack of every batch for X years or whatever.
I would suggest a retention plan that takes into account risk. What is risk? A consideration of the impact of a future problem and also its probability of occurring.
What are your risk factors? I'd say three that come to my mind are 1) the length you expect a customer to hold onto beer before consuming (this means a higher probability of a problem occurring) 2) the sugar content of the finished product and 3) the general complexity of the yeast in the recipe.
-
If you have an imperial stout, or a Christmas ale, or some other special, malt-forward beer, I think you can expect your customer to potentially hold onto it for a year or more. You might want to consider releases like this for a longer retention period in your "vault". Compare this to your pale ale that would probably be bland to gross twelve months later, with no other major seasonal milestone to save it for.
-
sugar. I'm sure you have heard of the juicy IPAs that customers forget to refrigerate and keep fermenting. Hopefully you'd have a big warning on the label, so maybe that would prevent the need to store these long term. Or any other beer that gets bottled at a relatively high specific gravity.
-
as you noted, the breaking down of complex sugars causing further fermentation is a risk. So any beer with a wild/saison/belgian yeast might have need for a longer retention period.
Perhaps, if you decide what storage capacity you can allocate, you would remove something according to these rules to allow room for new?
Or develop a program to sample the six pack throughout its retention period to evaluate how it ages? And therefore the risk of that beer existing in the wild?
Hopefully this wasn't condescending or stupid. Like I said, I'm applying general industry quality assurance principles to your problem.
Edit: further reading shows that there is research on disastatic yeast. Looks like it usually takes 25 days to bloom. Maybe you don't really need to keep samples more than a month or two. For any failure you observe, there should be research out there how to address/mitigate/contingency plan for that failure.