Skip to main content

Data Breaks

Scope​

The target is situations where, regarding data handled in software, states are observed where the "Source of Truth" cannot be uniquely determined.

Definition​

Data breaks refers to a state in which, regarding information that should originally be consistent, the location of the Source of Truth cannot be uniquely determined, and duplication, omission, and inconsistency are observed coexisting.

Symptoms​

  • Values with the same meaning exist in multiple places and do not match
  • Updated information is not reflected in specific screens or processing
  • Required data is missing, yet processing proceeds
  • Integrity violations are treated not as exceptions but as normal operations

Typical Triggers​

  • Copies and caches introduced as temporary measures are incorporated into operations as is
  • Assumptions made during migration are only partially shared
  • Situations are observed where data is rewritten through unexpected paths by external integration or manual operations
  • Situations are observed where reference and update points increase without the location of the source of truth being made explicit

Diagnostic Questions​

  • Is it a state where the source of truth of this value can be immediately answered?
  • Is it a state where data with the same meaning is updated in multiple places?
  • Is it a state where omission and inconsistency are assumptions rather than "exceptions"?
  • Is it a state where the scope of impact can be enumerated during fixes?

What This Is Not​

  • This does not discuss the appropriateness of normalization or denormalization
  • This is not a problem of choosing specific DB technologies or data stores
  • This does not refer to single input errors or accidental failures

Connections​

Last updated on by Hiroshi Kurabayashi