Storing data so it survives, and querying it so the answer arrives.
13 topics · 23 curated works
No prior grounding assumed.
A Simple Guide to Five Normal Forms in Relational Database Theory
William Kent · 1983
Explains the normal forms through the single informal principle that every non-key fact should depend on the key, the whole key, and nothing but the…
+1 more at this level
Assumes you know the vocabulary.
The Ubiquitous B-Tree
Douglas Comer · 1979
Surveys the B-tree and its variants as the reason so many storage systems converged on the same structure — one that keeps itself balanced and keeps…
+1 more at this level
Primary sources and full treatments.
A Relational Model of Data for Large Shared Data Banks
E.F. Codd · 1970
Proposes that data be organised around relations rather than pointers, so applications are insulated from how the data happens to be stored.
+18 more at this level
12 of 23 works
William Kent
Explains the normal forms through the single informal principle that every non-key fact should depend on the key, the whole key, and nothing but the key.
link checked 17 Sept 2026Zi Chong Kao
Teaches SQL entirely through one dataset of shipwrecks, running each query live in the browser, arguing the language is best learned by querying real data rather than reading syntax diagrams.
Douglas Comer
Surveys the B-tree and its variants as the reason so many storage systems converged on the same structure — one that keeps itself balanced and keeps disk accesses to a guaranteed minimum.
Surajit Chaudhuri & Umeshwar Dayal
Argues that analytical queries over historical data need a system modelled and indexed differently from the one that handles a business's live transactions.
link checked 17 Sept 2026E.F. Codd
Proposes that data be organised around relations rather than pointers, so applications are insulated from how the data happens to be stored.
Assumes basic set theory and predicate logic; no SQL, which did not exist yet and is not what this describes
11 pageslink checked 17 Sept 2026E.F. Codd
Introduces third normal form and the general argument for normalisation: that a relation should be decomposed until every non-key attribute depends on the whole key and nothing else.
Donald D. Chamberlin & Raymond F. Boyce
Proposes a query language built from English keywords over Codd's relational model, arguing a language readable by non-programmers would make relational databases usable outside a research lab.
Peter Pin-Shan Chen
Proposes modelling data as entities and the relationships between them, independent of whether the eventual database is relational, network or hierarchical, giving designers a notation to reason in before committing to an implementation.
Jim Gray, Raymond A. Lorie, Gianfranco R. Putzolu & Irving L. Traiger
Defines a hierarchy of lock granularities and named degrees of consistency weaker than full serialisability, giving system designers a vocabulary for the isolation trade-offs they were already making informally.
Selinger et al.
Gives System R's cost-based approach to picking a query plan — estimating the cost of each join order and access path and choosing the cheapest — the method essentially every relational optimiser still descends from.
E.F. Codd
Extends Codd's own relational model with entity types, roles and multiple relationships, addressing criticism that the flat original model lost meaning the real world actually has.
Jim Gray
Defines the transaction as the unit that must appear atomic, consistent, isolated and durable, and argues this one abstraction is what lets application programmers ignore concurrency and failure.
link checked 17 Sept 2026