Publications · Book, 129 pages, 2026
OntoDB: Database Internals by Example
TL;DR How a database engine works, from pages and B+ trees to query plans, inference, locking, logging, and ARIES recovery. OntoDB, a database management system for ontologies, is the running example.
Read the PDF Code Manual Lecture
Overview
Each chapter explains the theory first, then shows how OntoDB applies it, and lays out its design decisions: the choice, the alternatives, why, and what it costs. A list of all of them follows the contents, and every chapter ends with questions, answered in an appendix. Five parts follow a request through the engine: foundations; storage (pages, the buffer pool, B+ trees, indexing RDF, and log-structured trees for contrast); query processing (plans, the iterator model, joins, optimization, outer joins and aggregation, materialized RDFS inference with semi-naive evaluation and delete-and-rederive, and semantic search with interval labels and vector indexes); transactions (isolation and concurrency control); and recovery (the write-ahead rule, ARIES with a worked example, how OntoDB's persistent databases work today, moving the shell onto pages, and keeping the log small with truncation, full-page writes, and fuzzy checkpoints). Chapters for features not built yet say so. The book grows as OntoDB does.
Read
On a phone, open the PDF in your browser's reader.