Data Consistency
One of the companies I work for has millions of employees in total, and keeps records on them in several databases that exchange data through flat files. While this system seems to violate the principle of a single source of data, it’s been in place for years and works well. There must be a reason for this replication of data, and for allowing some inconsistency.
People almost always expect their cases to be unique and different. A person may fall behind on their insurance or their 401k contributions for reasons beyond their control, and they don’t expect to lose their benefits right away. The business rules may, ideally, require the employee's benefits or their company contributions to be cut off, but their manager may intentionally allow an exception to occur.
For this reason, it’s almost impossible for the data to be consistency managed in one central place. Instead, the company relies on a system of alerts that are triggered when what appears to be exception exists. The alert can always be ignored or overridden.
The photo above is of Houston, Texas, and is taken—like all photos on this website—with google maps.