> AI makes it much less painful to move off of legacy and/or overpriced software.

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.

▲

Thaxll 13 minutes ago | parent | next [-]

DynamoDB is very expensive.

▲

maerF0x0 2 hours ago | parent | prev | next [-]

DocumentDB more or less is MongoDB though

▲

richwater 3 hours ago | parent | prev [-]

DynamoDB is really a fantastic product for the 99% percentile of workloads that need a particular kind of super low latency read operation at large scale.

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.

	▲	joshstrange 3 hours ago \| parent [-]
		If your access patterns are fixed and not going to change, if your schema is pretty much solidified, then yes, I don't doubt that at all. But few things I've worked on have had such stable access patterns or schemas that they would benefit. I enjoyed solving the challenges around using DynamoDB way more than I enjoyed using it on an ongoing basis.