It is possible to distribute a logically coherent set of data across several BS2000 servers in a computer network. The benefit of using several database servers is that the load of each server is significantly lower than the load produced when using a single database server. Equally, distributed databases limit the maximum damage which can result from the failure of a single database server.
Database distribution is controlled by tables. These can be modified at any time, thus enabling distribution to be adapted dynamically to react to changing situations.
Distributed processing can be started and terminated separately from the independent DBH session. In particular, several UDS-D sessions can be started sequentially in a single independent DBH session, using the DAL commands &START and &CLOSE.
The administrative handling of transactions in PTC status ("prepare to commit") can be modified dynamically. This guarantees completely autonomous administration ("site autonomy"), even in the event of global faults or of a limited availability of network components (see section “Modifying PTCSYNCH values (MODIFY PTCSYNCH)”).