Distributed systems like mongodb or microservices, are often considered “scalable”, whereas non-distributed systems (postgres without sharding) are not.
It seems like we’ve developed a collective bias to always pick these “scalable” tools and my theory is that this is fear-driven development.
New job. 20+ microservices, mongo, postgres, rabbitmq, elastic, kubernetes cluster. $50k/year+ to run, averaging <100 active users. Peaks causing performance problems.
I learned then that it’s not the tools you have, it’s how you use them. The self-inflicted DDOS-ing could’ve been prevented, and the solution was not mongo.