Database Internals & Performance Engineering

How a relational database splits memory into shared and private regions, and why that split explains almost every performance behavior you'll see later — using Oracle's SGA/PGA model as the clearest teaching example.

How a table's rows actually end up laid out on disk — the block, extent, and segment hierarchy that underlies every I/O decision an optimizer ever makes.

Why one line of coding habit — using bind variables instead of literal values — is the single biggest lever over both query performance and SQL injection safety.

The mechanics behind two guarantees you've likely taken for granted: that a rollback can undo your changes, and that a committed transaction survives a crash.

The methodology behind tuning: where to actually look first, how statistics feed the optimizer's decisions, and the ideas of selectivity and cardinality that explain why the same table gets different plans for different queries.

How to generate an execution plan and actually read it — cost, row estimates, join order, access paths, and the specific warning signs that separate a healthy plan from a dangerous one.

When a hint is actually the right tool versus a sign you're fighting the optimizer, plus the rewriting techniques that usually fix a slow query without needing a hint at all.

The real story behind "why is this query slow right now" — database health reporting, wait events, and how to diagnose lock contention while it's actively happening.