Oracle Distributed Systems

Communication Flow

Yourreplicated environment is probably not as simple as a master site and a snapshot site. Oracle allows you to mix and match endless permutations of masters and snapshots. A site can even be a snapshot site for one set of tables and a master site for another. Somehow, all of the data gets to its destination.

The most complex configuration for updateable snapshots is one in which two or more multi-master sites have their own snapshot sites, as shown in Figure 13.1.

Figure 13-1. Updateable snapshots where multi-masters have their own snapshot sites

Suppose that a user at the snapshot site PSLS.BIGWHEEL.COM makes an update to the SALES_LEADS table. How does that change propagate to the updateable snapshot of the same name at site PSFO.BIGWHEEL.COM? The process is as follows:

1. The user updates the snapshot SALES_LEADS.

2. The trigger USTRG$_SALES_LEADS creates an entry in the snapshot log USLOG$_SALES_LEADS. (In Oracle8, this trigger is internalized and therefore not visible in the data dictionary.)

3. The snapshot refresh updates master table in PHQS.BIGWHEEL.COM.

4. The replication trigger SALES_LEADS$RT in PHQS.BIGWHEEL.COM queues the update for propagation to its companion master, PCAL.BIGWHEEL.COM. (In Oracle8, this trigger is internalized and therefore not visible in the data dictionary.)

5. When the update is applied at PCAL.BIGWHEEL.COM, the trigger TLOG$_SALES_LEADS creates an ...