Learning Pathway
06 — Managed PostgreSQL in the Cloud: Architecture & Trade-offs
16 lessons- 01 What “Managed PostgreSQL” Actually Means: The Shared Responsibility Model What a managed PostgreSQL service runs for you and what you still own — the control plane vs data plane split, and the shared…
- 02 Two Architectures: “Postgres-on-a-Disk” vs Cloud-Native Disaggregated Storage The single most important distinction in managed PostgreSQL: classic instance-plus-volume services (RDS, Cloud SQL, Flexible Server) versus disaggregated, log-as-the-database designs (Aurora, AlloyDB, Neon, HorizonDB).
- 03 AWS RDS for PostgreSQL: The Managed Baseline How Amazon RDS for PostgreSQL works: community Postgres on EBS, Multi-AZ synchronous standbys, Multi-AZ DB clusters with readable replicas, automated backups and the limits…
- 04 AWS Aurora PostgreSQL: The Log-Is-the-Database Storage Engine How Amazon Aurora PostgreSQL re-architects storage: shipping only redo log to a distributed six-copy storage fleet across three AZs, shared-storage replicas, fast failover and…
- 05 Google Cloud SQL for PostgreSQL: Google’s Managed Baseline How Cloud SQL for PostgreSQL works: community Postgres on Persistent Disk, regional HA with synchronous disk replication, read replicas, automated backups and the Enterprise…
- 06 Google AlloyDB: Disaggregated Storage and a Columnar HTAP Engine How AlloyDB re-architects PostgreSQL: a log-processing storage service that materializes pages, read pools that share storage, and an in-memory columnar engine for analytics on…
- 07 Azure Database for PostgreSQL Flexible Server: Community Postgres on Azure How Azure Database for PostgreSQL Flexible Server works: community Postgres on Azure managed disks, zone-redundant synchronous HA, async read replicas, a built-in PgBouncer and…
- 08 Azure HorizonDB: Microsoft’s Cloud-Native Distributed PostgreSQL What Azure HorizonDB is: Microsoft's cloud-native, PostgreSQL-compatible database with disaggregated storage, elastic compute and AI integration — how it fits the log-is-the-database family and…
- 09 Neon: Serverless PostgreSQL, Scale-to-Zero, and Copy-on-Write Branching How Neon re-architects PostgreSQL as a serverless service: stateless compute, a storage layer of pageservers and Paxos-backed safekeepers, bottomless object storage, scale-to-zero and instant…
- 10 Databricks Lakebase: Lakehouse-Native PostgreSQL OLTP What Databricks Lakebase is: a managed, Neon-based serverless PostgreSQL OLTP database fused with the lakehouse — separation of compute and storage, branching, and built-in…
- 11 High Availability & Failover Models Compared How HA really works across managed PostgreSQL: synchronous standbys on a disk (RDS, Cloud SQL, Flexible Server) versus shared-storage failover (Aurora, AlloyDB) versus consensus…
- 12 Read Scaling: Read Replicas vs Shared-Storage Replicas vs Read Pools How to add read capacity in managed PostgreSQL: asynchronous streaming read replicas (RDS, Cloud SQL, Flexible Server) versus shared-storage replicas and read pools (Aurora,…
- 13 Backups, PITR, and the Durability You Actually Get How automated backups and point-in-time recovery really work in managed PostgreSQL: snapshots plus continuous WAL archiving, retention windows, restore-to-new-instance, and why HA is not…
- 14 Connections and Pooling: Why a Pooler Is Almost Mandatory Why PostgreSQL's process-per-connection model forces connection pooling in managed environments, and how built-in poolers (RDS Proxy, Azure PgBouncer, Neon proxy, Supavisor) and serverless workloads…
- 15 What You Give Up: No Superuser, Extension Allowlists, Locked GUCs, and Lock-In The real constraints of managed PostgreSQL: losing superuser, extension allowlists, parameters you cannot change, the absence of OS access, and how cloud-native forks deepen…
- 16 Choosing a Managed PostgreSQL: A Decision Framework and Cost Model A practical framework for choosing among RDS, Aurora, Cloud SQL, AlloyDB, Azure Flexible Server, HorizonDB, Neon and Lakebase — driven by workload, scale, cloud,…