In order to improve the behavior and performance of UDS/SQL, it is essential that the databases being used are structured appropriately for the applications involved. A number of facilities are available in UDS/SQL both for the logical and the physical organization of data, and thus for optimizing access to data stored in the database (see the "Design and Definition" manual for details).
One method of increasing performance substantially is to avoid unnecessary I/O operations when accessing data stored in databases. A description of how I/O behavior can be analyzed and controlled can be found in section “Optimizing I/O behavior”.
The amount of working memory required for the UDS/SQL configuration can be reduced by the concurrent sharing of programs that are loaded as subsystems. A description of the various subsystems that can be created for UDS/SQL is provided in section “Optimizing usage of working memory with the subsystem functionality of UDS/SQL”.
In the case of multi-processor systems, the performance of the independent DBH can be improved dramatically by the dynamic distribution of UDS/SQL tasks to different processors. The methods used to analyze and control processor utilization are dealt with in section “Optimizing processor utilization with the independent DBH”.