Chapter 15. Runtime Updates

Data change events are used by JBoss Data Virtualization to invalidate resultset cache entries. Resultset cache entires are tracked by the tables that contributed to their results. By default JBoss Data Virtualization will capture internal data events against physical sources and distribute them across the cluster. This approach has a couple of limitations. First, updates are scoped only to their originating VDB/version. Second, updates made outside of JBoss Data Virtualization are not captured. To increase data consistency, external change data capture tools can be used to send events to JBoss Data Virtualization. From within a cluster the org.teiid.events.EventDistributorFactory and org.teiid.events.EventDistributor can be used to distribute change events. The EventDistributorFactory can be looked up by its name "teiid/event-distributor-factory". See the example below.

InitialContext ctx = new InitialContext();
EventDistributorFactory edf = (EventDistributorFactory)ctx.lookup("teiid/event-distributor-factory");
EventDistributor ed = edf.getEventDistributor();
ed.dataModification(vdbName, vdbVersion, schema, tableName);

This will distribute a change event for schema.tableName in VDB vdbName.vdbVersion.

When externally capturing all update events, the "detect-change-events" property in the "teiid" subsystem can be set to false, so change events will not be duplicated. By default, this property is set to true.

Use of other EventDistributor methods to manually distribute other events is not always necessary. See System Procedures in Red Hat JBoss Development Guide: Reference Material for SQL based updates.