From OpenSCADAWiki
In the case of the "Data acquisition" subsystem, data reservation (Fig. 12) serves as:
- Redundancy of the data acquisition mechanism. Typically, this function is realized without special arrangements by simply running of the parallel redundancy stations with the same configuration and working independently. However, in the case at the station working as PLC, such approach is unacceptable because of the simultaneous making of the control actions and the absence of synchronization of data of the calculators.
- Compensation of the data loss on the time of the node inaccessibility with the redundant node archive. There are two mechanisms of the compensation. The first and the main mechanism implements the loading of the sections of the archive from the redundant station at the time of the station startup in general or of individual controller objects. The section of the archive is requested from the moment of the last record in the local archive and till the current time. The depth of the request is limited by the indicating of the limit time in the configuration of the redundancy. The second, complementary mechanism, performs the filling of the "holes" in the archive at the time of the actual user's request to the data. Such an approach on the one hand allows to make the predictable in time synchronization at startup and on the other hand it's actually eliminates the data loss in the case of working at least one station during the entire time.
- Distribution of the data acquisition load between the nodes. When creating complex allocated systems there can be an important question of predicting and optimizing of the overall system performance. Taking in account these problems, the redundancy mechanism provides execution of the data acquisition tasks of individual sources (OpenSCADA controller objects) only at one station. The other stations' tasks would go to the data synchronization mode with the executive station. In the case of loss of the connection with the executive station, the local data acquisition task is started. It is also provided for the possibility of an optimal load distribution for executing the tasks of the data acquisition for a group of controller objects, between the stations.
- Optimization of the load on the external data sources through the data request from an external source by the only one node. In practice, we often meet highly loaded data sources or interfaces of accessing to the data sources, for which even the data acquisition by one station can be a problem and would require reducing the acquisition periodicity, i.e. data quality. The mechanism of redundancy, besides the load distribution between the stations as described above, allows you to remove additional load form the data source and its interfaces, thereby improving the quality of the data. Writing to attributes of the controller object under redundancy causes to the modification request sending to the master one, that is — through it.
- Prevention of some differences of the data on different nodes associated with the mismatch of moments of time at the independent acquisition of the data by individual nodes, performs in a way of receiving the data from the station with the active controller object. In systems of high accountability under the redundancy there should be excluded or minimized the differences in the data at different stations, that means the real acquisition of the data by one station and synchronization with these data of other ones.