Why I still choose PostgreSQL for large-scale systems.
PostgreSQL continues to be one of the best choices for modern backend systems.
Reasons:
excellent reliability
strong transactional guarantees
indexing flexibility
scalability options
ecosystem maturity
strong observability tooling
...
Kafka becomes powerful when retries are designed correctly.
A lot of teams use Kafka for scalability.
But reliability is where event-driven systems actually become difficult.
Important things that matter in production:
idempotent consumers
retry handling
dead-letter queues
ordering...
Most backend systems scale the wrong layer first.
A lot of engineering teams immediately add more servers when performance drops.
But in many production systems, the actual bottleneck is:
inefficient database queries
poor caching strategy
synchronous workflows
missing indexes
...