|
|
|
|
In the 1990s, active features in database systems have become common place. Most commercial systems now support some kind of trigger processing. A scalable trigger system is an active database that can support multiple triggers on a single data element. Although trigger support has become more popular, scalability in these systems is non-existent. From an application point of view, scalability allows for the implementation of workflow modeling in an active database system. Scalable trigger support allows for workflow diagrams to be compiled and implemented by the active database system. Several commercial products already do this, but are at the application level. These workflow diagrams would run substantially faster inside an active database. There are several issues facing scalability being investigated at OU. Some issues include trigger priority and selection during execution such as whether they can be checked in parallel. Another is determining the capacity of active (running) triggers in the system. Some triggers are used more frequently than others are, so trigger-caching strategies must be implemented. Semantic analysis of triggers prior to execution can be done to minimize the number of them in a system. Composite event handling must also be examined for possible performance benefits. Also, predictability of trigger execution can be a great tool for optimization. Strategies for synchronous as well as asynchronous trigger support must be investigated. In addition, recovery schemes for scalable trigger systems will need to be rethought. To solve these issues, a conceptual model for scalability trigger support will first be constructed. The plan is to use Informix as the host database system and construct a scalable trigger support module with its datablade API package.
|
|
For problems or questions regarding this web contact database@cs.ou.edu. |
