What are the risks involved in a DIY solution like RODA?

Question

I am curious to hear from anyone, particularly smaller institutions, that have implemented DIY solutions with Archivematica, RODA, etc. We have a non-trivial amount of data (300TB - 400TB) that would be cost prohibitive in vendor-based pure cloud scenario using S3 and Glacier.

It seems to me the most cost effective method is to have data on-site in duplicate and replicated to Amazon Glacier, perhaps using a commercial system for the added support. However, its seems (in my ambitious mind) possible to manage a system like RODA in-house and handle the replication ourselves, needing only to pull from Glacier if the fixity checking reports an error. Can someone disabuse me of my fanciful notions? Any horror stories or success stories? Am I right to assume that as long as the bags are stored in Glacier we could migrate those to another system later on if we found maintaining our own system too unwieldy? How horrifying is it to all of you to rely on Glacier as a fail safe option?

Comments

This tweet I just saw https://twitter.com/timgollins/status/1095731181604298752 reminded me of your question here.
I don't have a useful answer for you as I've not done what you are suggesting. However I would like to convey my concern that you believe that using e.g. S3 or glacier are cost prohibitive for you. That, to me, is an indicator that perhaps your organization cannot afford to do digital preservation and should instead find a way to ensure your assets/artifacts are preserved by e.g. raising more funds, or donating them to an organization that has the capacity to adequately steward them.