Agreed. While DynamoDB != MongoDB, they are similar. I built a product on DynamoDB (Single-table-design) and while it was cool to find a way to make relational data work in DynamoDB, it was always a chore. New feature development was a huge PITA as was changing schema in any way. Sure, I might have been "holding it wrong" but after a weekend with Claude I migrated everything to Postgres and have been incredibly happy with that change something like 4-6 months later.
The simplicity of using it and the lack of operations is really why most people tend to reach for it over a relational option, although over the years the burden of managing an RDMS has consistently gone down.
I enjoyed solving the challenges around using DynamoDB way more than I enjoyed using it on an ongoing basis.
Oracle has understood this for a long time. RDMBS have seen many generations of application tech stacks rise and fall. Other than C and unix they have outlasted almost everything.
They bought Cerner early in the decade and have whittled it down to absolutely nothing. This is a problem when the software that Cerner sold was EMRs and I have people who used to work there telling me that the perioperative and anesthesia suites now have one person working on them, and no one's fixing issues.
They're a hypercloud provider now. It may be awhile.
Ehhhhhh maybe https://www.cnbc.com/quotes/ORCL
As much as I'd like to believe this, they have their ways of making money.
it's end user software that's absolutely getting crushed.
AI greatly removes that differentiator. AI doesn’t care if it needs schemas or if there’s an idiomatic driver for Python or Node. Even schema changes aren’t an issue and most databases support complex data anyways.