@sapphirescales Your math is based on faulty premise.
Inefficiencies amplify with scale. The data I posted above was from a simple all text just a bit more than 1 page document. A 5x storage increase in that file is negligible, from 100k to 500k. BUT... I'm not working with that simple of a document am I?
Oh no... I'm dealing with in this specific case, full mortgage packages. Which takes a file from 10mb to 150mb. This results in 100gb of document creep PER YEAR. When the platform is holding almost 20 years of loans, it adds up.
I mean you're still not wrong, I've got a server with 1.5tb of SSD storage, that's a bit more than 5 years old. The new platform simply needs 3 TB of storage and it's good for the lifespan of the replacement server. We plan for these things.
But that doesn't make the usage efficient, which is why I'm working with the client to determine a proper data lifecycle. Which means after x years, I'm going to run this script on the folder that passes each PDF through ghostscript to lower quality settings to 150dpi, which in general cuts all the sizes to a quarter or less of what they are currently.
If they use proper methodology generating their PDFs that compression process is rendered useless, because they could indeed store many millions more of documents. But it seems to me that you're having trouble imagining a business at scale that can actually generate data like this. Which is pretty silly considering I'm watching it happen at an office that owns 30 desktops. They aren't exactly huge.
Also, cloud migration means paying per GB of storage, that's a direct monthly cost that needs containment. Furthermore every GB of data you have to restore during an outage is that much more time you're down costing even more money.