Umair Shahid, Stormatics Pte Ltd
This presentation talks about high availability for PostgreSQL by treating it as a layered set of replication and failover patterns. Many teams deploy synchronous replication without understanding its trade‑offs or rely on single‑region clusters that can’t survive a data‑center outage. Worse, budget constraints often force compromises that jeopardise recovery objectives. The talk proposes a framework for composing HA layers - local synchronous replication, asynchronous standbys, hot standbys and cross‑region disaster‑recovery nodes - to meet specific RPO/RTO targets while controlling costs. It will cover practical experiences from multiple deployments where achieving four-nines availability was critical, yet enterprise clustering licences were not an option.
Umair Shahid is the founder of Stormatics, a PostgreSQL-focused professional services firm that designs, operates, and scales production database platforms for fintechs, SaaS companies, and large enterprises. He works on high availability architectures, performance optimisation, database security, and migrations from expensive proprietary databases to PostgreSQL, both on premises and in the cloud.
Umair is a veteran of the PostgreSQL community. Being a recognized subject matter expert, he is a regular speaker at conferences globally and publishes blogs that get 15,000+ visitors each month.
author = {Umair Shahid},
title = {Designing Layered {High-Availability} Architectures for {PostgreSQL} on a Budget},
year = {2026},
address = {Seattle, WA},
publisher = {USENIX Association},
month = mar
}
