Uploaded June 2026 | Updated September 2026, 2 weeks ago
In Part 1, Rick Houlihan shows how to read any modeling decision as a setting on three dials: pattern complexity, read/write mix, and the size of a write versus the size of a read. Turn the dials and you can predict which pattern pays and which one quietly imports write amplification or slams into a bounded read path.
What you'll learn:
- The three dials that decide a data model — how to read them before you write a line of schema
- Why copy-on-write storage turns "just embed it" into write amplification, and how the 16 MB document and the 100 MB aggregation ceiling shape the patterns
- Where each classic document pattern pays off — and where the bet inverts
- When single-collection / single-table design wins, and what breaks it
- The 90/10 problem: why a single-model engine always fails one side of the line
Part 1 of 2. Part 2 shows what changes when the join is free. Live demo on Oracle 26ai + Q&A
In Part 1, Rick Houlihan shows how to read any modeling decision as a setting on three dials: pattern complexity, read/write mix, and the size of a write versus the size of a read. Turn the dials and you can predict which pattern pays and which one quietly imports write amplification or slams into a bounded read path.
What you'll learn:
- The three dials that decide a data model — how to read them before you write a line of schema
- Why copy-on-write storage turns "just embed it" into write amplification, and how the 16 MB document and the 100 MB aggregation ceiling shape the patterns
- Where each classic document pattern pays off — and where the bet inverts
- When single-collection / single-table design wins, and what breaks it
- The 90/10 problem: why a single-model engine always fails one side of the line
Part 1 of 2. Part 2 shows what changes when the join is free. Live demo on Oracle 26ai + Q&A
