Release 10.1C: OpenEdge Replication:
User Guide
Using synchronous or asynchronous replication
OpenEdge Replication supports two methods of replication: synchronous and asynchronous. Figure 1–6 shows both synchronous and asynchronous methods of replication.
Figure 1–6: Synchronous and asynchronous configurations
![]()
In Figure 1–6, the asynchronous replication supports a maximum of two OpenEdge Replication agents and the synchronous replication supports only one OpenEdge Replication agent. There is one OpenEdge Replication agent for every target database.
During asynchronous operation, the user changes records and the transactions are committed without acknowledgement and sent back to the OpenEdge Replication server. Without waiting, the OpenEdge Replication server sends more AI blocks from the AI transaction log to the OpenEdge Replication agent, and the OpenEdge Replication agent applies these changes to the target database. Of the two configurations (synchronous and asynchronous), asynchronous performs better.
Figure 1–7 shows asynchronous operation in the OpenEdge Replication model.
Figure 1–7: Asynchronous operation
![]()
During synchronous connection, the user changes records and the transactions are committed. When the OpenEdge Replication agent encounters a transaction end, it sends an acknowledgment back to the OpenEdge Replication server. The committing user will block (wait) until the transaction is fully applied to the target database. Other users are not blocked during this activity.
Of the two configurations, synchronous is the safest; however, it is also a low-performance option. For more information on choosing asynchronous versus synchronous mode, see the "Choosing a hot standby database" section.
Figure 1–8 shows synchronous operation in the OpenEdge Replication model.
Figure 1–8: Synchronous operation
![]()
Because of the user blocks in the synchronous model, performance will be much slower than in the asynchronous model.
Copyright © 2008 Progress Software Corporation www.progress.com |
![]() |