This section describes possible data loss using the high-availability services:

Updates to objects only lose uncommitted creates, updates, and deletes to synchronous replication nodes when an active node fails. No application committed work can be lost.

Updates to objects using asynchronous replication lose any queued creates, updates, and deletes when an active node fails. Committed application work can be lost.

Inbound communications buffers that have not been processed by ActiveSpaces® Transactions are lost. Some network buffers are configurable. A smaller network buffer size implies lower exposure to data loss.

You can avoid even this small risk of data loss if the client of the ActiveSpaces® Transactions application has a protocol acknowledgement and includes retry logic if no acknowledgement is received. If the client application resends the request nothing is lost. However, care should be taken to handle duplicates.