PostgreSQL performance issues often show up first in the application: latency spikes, connection exhaustion, and unpredictable throughput under load. Because PostgreSQL connections are relatively expensive, “one request = one new connection” fails quickly at modern concurrency levels — across every language and framework.
This talk focuses on practical, cross-stack patterns that reduce overhead without rewriting your system. We’ll start with round-trip reduction (batching statements, using prepared statements effectively, and avoiding chatty ORM behavior), then move to application-level pooling patterns in common ecosystems (Java, .NET, Python, Node) and how to size pools sensibly.
Finally, we’ll connect the dots to production architecture: when client-side pools are enough, when you should introduce server side pooling like PgBouncer, and how to avoid the most common failure modes (stale connections, idle-in-transaction, pool storms). You’ll leave with a concrete checklist you can apply to any PostgreSQL-backed service.