I disagree, it’s quite performant and cost effective. I work for a very large software company and it scales to our needs very well. However, it is emphatically not the solution to everything nor is it a replacement for all traditional RDBMS use cases. It also takes a quite different mindset when thinking about your data than when using SQL.
IMO what it does really well is handle being a persistent data store for well-formed REST API endpoints. When you understand your access patterns and implement your GSIs correctly, a RDBMS can’t match the performance even with well tuned queries. Dynamo excels at giving you a record set when you know exactly which set of records you want and it’s based on one or two very simple conditions.
Where it falls behind is for data warehousing and reporting use cases. Dynamo is comparatively slow and inefficient when it comes to asking complex questions about the data. RDBMS systems are built for that use case and as such have extensive tools to optimize whatever wild queries you want to throw at it.
If you’re interested in learning about single table design, which is not good for all cases, check out this video. I’ve watched it quite a few times and it’s been the biggest help in wrapping my head around how to do the data modeling for it. youtu.be/KYy8X8t4MB8
Ok my steam deck finally finished updating. Time to go. 😁