One of the most important concepts of UDS/SQL is that it distributes data from one database across several files, known as realms. This increases the availability of data managed by UDS/SQL, since the effects of device, write or read errors are localized to subunits in the database.
Utility routines do not always need to use all realms to perform updates. This minimizes the potential restrictions of parallel DBH operation.
Copies of databases can be created on several local or remote servers. Replicates are updated with the ALOG files of the original database so that, if the original data becomes unavailable, backup data can quickly be accessed and processed. A further advantage of this distributed data management is that spatially distributed read access can also be accelerated by the replicas (local storage concept).