 Linter Multiversion is a DBMS supporting the multi-version data processing model.
As the information scope is constantly growing, parallel processing of concurrent transactions is more and more in demand. Linter Multiversion is intended for developing applications requiring conflict-free read and write queries in a multi-user system. Linter Multiversion is recommended for applications in which read and write query conflicts substantially affect the system's performance if a classic blocking database server is used.
Linter Multiversion uses the powerful functionality of the basic Linter Standard version. The system provides high availability of data and faster processing of concurrent transactions thanks to multi-version data storage supporting the ANSI levels of transaction isolation.
Linter Multiversion allows the developer to create applications for the following operating systems:
The multi-version capabilities make data viewing and data modification independent processes, since modifying a row in a table leads to creating a new version of this row. Generally, a special process of "cleaning" outdated versions is not required since they are reused by the system.
The multi-version mechanism improves the paralleling in the multi-user mode. The system also supports the standard transaction modes, from Dirty Read to Serializable.
Thanks to the multi-version mechanism, each application task can work with its own version of initial data independently for a long time. Synchronization of changes made to the same data by different programs is required only when committing a transaction.
If two tasks have modified the same data, successful committing of the modification is possible only for one of them. Modifications made by the other task will not be committed.
Developers should take into account that the multi-version mechanism itself requires additional resources. If the server workload (number of parallel transactions) increases, the performance of the multi-version mechanism starts falling due to frequent updates necessary to support multiple data versions. |
