AI Glossary / Data engineering and architecture / Intermediate
Codd's 12 rules
Definition
Codd's 12 rules are criteria proposed by relational-database pioneer Edgar F. Codd to assess whether a database management system genuinely supports the relational approach. The familiar name includes twelve numbered rules plus a foundational Rule 0. They emphasize logical access, integrity and independence from physical implementation rather than merely displaying data in tables.
What it means
The rules arose in the context of product claims about being relational. They provide a conceptual benchmark, not a modern certification automatically awarded to every SQL database. A system can support useful relational capabilities while differing from the full ideal. The rules also address the database management system, not just whether an individual table has a primary key. They help users ask what remains stable when storage, schema or deployment details change.
How it works
The foundation requires management through relational capabilities. The twelve themes are: tabular information; guaranteed logical access; consistent null handling; a queryable relational catalog; a comprehensive data language; updates through theoretically updatable views; set-level insertion, updating and deletion; physical independence; logical independence; independently declared integrity; distribution independence; and protection against bypassing constraints through lower-level interfaces. Together they connect data representation with manipulation and protection. Reading the original wording matters when evaluating a particular rule because a short theme label omits qualifications.
Three practical examples
- A developer expects a query to keep working after an administrator adds an index, illustrating separation from physical access methods.
- A team declares a relationship constraint in the database so several applications share the same enforcement rather than each reimplementing it differently.
- An analyst queries a system catalog to inspect table and column definitions using the database's supported logical interface.
Think of it this way
Think of a building standard that evaluates not only whether a structure has rooms, but whether its wiring, access and protections work coherently. A table-shaped interface alone does not establish the deeper properties. The analogy is limited because Codd's rules are a historical conceptual benchmark for relational systems, not a building code or a current legal certification scheme.
Worked example
Suppose an Orders table contains customer identifier C17, while Customers contains only C10 and C11. A declared foreign-key constraint can reject that order because the referenced customer is absent. If one application checks this relationship but a bulk-loading path bypasses it, the system can still admit inconsistent data. The example illustrates why integrity should not depend solely on one application's interface and why bypass paths matter. It does not establish compliance with every Codd rule: view updating, logical independence and the other criteria require their own analysis.
Illustrative case study
Fictional case: Maple Records evaluates a database migration. The team initially asks only whether the target accepts its SQL syntax. It then checks catalog access, constraint enforcement, update behavior and whether physical tuning requires application changes. A few differences require changes to its migration plan. The team describes those capabilities directly rather than claiming complete rule compliance from one successful query. This fictional case illustrates a structured comparison, not an assessment of any named product.
Common misconceptions and mistakes
Do not say there are only twelve numbered items if including the foundation: numbering runs from 0 through 12. Do not equate SQL support with complete adherence to the relational model. The view-update rule does not mean every imaginable view can be updated unambiguously. Physical independence concerns logical application behavior, not a promise that performance never changes. Also, the rules are not the same as normalization forms or the ACID transaction properties, although these topics relate to database quality.
Business use
Businesses can use the principles to ask practical questions about portability, integrity and maintenance. Which changes force application rewrites? Where are constraints defined? Can administrative or loading tools bypass them? How can metadata be inspected? Evaluate these questions against actual application needs and documented product behavior. For AI data platforms, shared constraints and clear metadata can help preserve trustworthy inputs, but they do not establish that labels or source observations are factually correct.
People also ask
Why are there thirteen numbers in the twelve rules?
Rule 0 is the foundation accompanying the twelve main rules.
Are these the same as database normalization?
No. Normalization concerns schema design; the rules describe broader capabilities of a relational DBMS.
Does a table-based product automatically satisfy them?
No. Representation alone does not establish manipulation, independence or integrity behavior.
Check your understanding
Adding an index changes query speed but leaves the application's logical queries and results intact. Which principle does this illustrate?
Answer: Physical data independence. The physical access arrangement changed without requiring a logical application rewrite; performance equality is not the claim.
Related terms
Learn next
Select three rules relevant to an application and turn each into a concrete test or documentation question. For example, inspect metadata, test a rejected invalid reference and add an index without changing query logic. Avoid treating these limited checks as proof of complete relational compliance.
More in Data engineering and architecture
- Cluster hypothesis - The cluster hypothesis in information retrieval is the idea that documents that are closely associated tend to be relevant to the same information needs.
- Clustered file system - A clustered file system allows multiple computers in a coordinated cluster to access a shared file-system namespace while managing concurrent access and...
- Coalescence Rule - The coalescence rule is a database dependency inference rule that combines a multivalued dependency with a suitably related functional dependency to...
- Cognition enhanced Natural language Information Analysis Method - Cognition enhanced Natural language Information Analysis Method, usually abbreviated CogNIAM, is a named method for modeling organisational knowledge.
- Cognitive map - A cognitive map is an internal representation of relationships in an environment that helps a person or animal orient, reason or plan.
- Collective operation - A collective operation is a communication operation involving a defined group of processes in a parallel program.
Sources
Primary text: Codd, Is Your DBMS Really Relational?, preserved transcription, https://thaumatorium.com/articles/the-papers-of-ef-the-coddfather-codd/1985a-is-your-dbms-really-relational/ . Supporting primary vendor manual: IBM SQL/400 implementation guide, https://www.bitsavers.org/pdf/ibm/as400/GG24-3321-01_SQL400_-_Guide_for_Implementation_OS400_V1R3M0_199008.pdf . Original-text transcription opened; archival fidelity and product-specific editorial review remain pending.
Previous: Cochran's Q test Next: Code-break procedure