Part 2: Engine Deep Dives

"Every database hides a machine underneath. The engineers who win are the ones who know the machine."

Overview

Four chapters that go inside the most important engines in production today:

#	Chapter	Core Concept	Difficulty	Read Time
05	PostgreSQL in Production	WAL, VACUUM, connection pooling, extensions	Intermediate	45 min
06	MySQL & Distributed SQL	InnoDB, Vitess, NewSQL trade-offs	Intermediate	40 min
07	NoSQL at Scale	DynamoDB, Cassandra, MongoDB, Redis patterns	Advanced	50 min
08	Specialized Databases	Time-series, search, graph, vector engines	Advanced	45 min

Key Themes

• Internals matter for tuning — you cannot size shared_buffers correctly without knowing how PostgreSQL uses it
• Schema design is engine-specific — a DynamoDB single-table design would be catastrophic in PostgreSQL, and vice versa
• NewSQL is not a free upgrade — distributed SQL adds 10–100ms of consensus latency you cannot optimize away
• Specialization wins at the extremes — beyond a certain scale, a general-purpose database is the wrong tool

What You'll Be Able to Do

After completing Part 2 you will be able to:

• Tune a PostgreSQL configuration for a 64GB production server with annotated reasoning
• Explain why PgBouncer transaction mode breaks LISTEN/NOTIFY and prepared statements
• Design a DynamoDB single-table schema for an e-commerce access pattern
• Choose between ClickHouse, TimescaleDB, and InfluxDB for a given ingestion rate
• Explain when Neo4j traverses relationships faster than a SQL self-join

Prerequisites

Complete Part 1 — Foundations before starting here. The WAL mechanics from Ch01, the indexing concepts from Ch03, and the MVCC model from Ch04 are all assumed knowledge.

Continue to Part 3

After finishing these four chapters, continue to Part 3 — Operations for replication, sharding, and distributed transactions.