Data Integrity

The Department of Defense’s standard definition of data integrity (taken from Department of Defense Definition 5200.28, “Security Requirements for Automated Information Systems (AISs)”) is clunky but instructive: “the state that exists when data is unchanged from its source and accidentally or maliciously has not been modified, altered, or destroyed.” This is pretty straightforward: a secure system that preserves data integrity keeps the original data safe from tampering or accidental, unintended modification.

This requirement breaks neatly into two separate subrequirements. One is that a system with good data integrity controls protects data from modifications requested by users; most systems depend on access controls to deliver this property. The other, which is of greater interest to us, is that an integrity-preserving system should guard against (or at least warn us about) corruption of its data caused by hardware or software problems. This is particularly important for transactional database systems like Exchange and Microsoft SQL Server, because the loss or damage of a single transaction’s worth of data can have significant undesirable consequences.