Postgres isn’t slow, your storage is | POSETTE: An Event for Postgres 2026

This technical talk by Sai Srirampur (ClickHouse) demonstrates that many Postgres scaling issues with TB-scale datasets are actually storage I/O bottlenecks, not Postgres limitations. Through perf, flamegraphs, and Postgres internal stats, he shows how WAL fsync, buffer reads, and checkpoints dominate on slow storage but largely disappear on NVMe. Ideal for DBAs and engineers troubleshooting Postgres performance at scale.