Synchronizing Extensive Changes

Learn about propagating changes on standby nodes and remote restores.

As discussed in Data Synchronization, remote restores and ongoing updates are the two basic mechanisms the Vocera Voice Server uses to synchronize data among cluster nodes. The Vocera Voice Server performs this synchronization automatically; however, there are times when you may want to force the standby nodes to perform a remote restore for the following reasons:

Ongoing updates are effective for propagating incremental changes to standby nodes.

For example, adding and removing users, changing group permissions, and so forth.
Remote restores are more efficient for synchronizing large sets of changes to the database.

For example, importing 5,000 entities, updating users with a spreadsheet, transferring site data, and so forth.

The remote restore is effective for large sets of changes because it reads the entire database of the active node into a standby node's memory in a single operation, and then writes it to the standby node's disk. The ongoing updates process treats every edit as a separate operation.

Tip: Use the remote restore mechanism to synchronize large sets of changes. Using remote restore avoids having the standby nodes out-of-sync for long periods of time, and it avoids creating excessive network traffic.

To use a remote restore to force synchronization to occur:

Back up your database.
This action also causes standby nodes to perform a backup.
After the backup completes, stop the standby nodes.
Perform all the necessary updates on the active node.

Note: The active node remains running, and the badges remain connected to it.
Restart the standby nodes.
The standby nodes automatically perform a remote restore, synchronizing data quickly.

Similarly, when you first set up a cluster, the best practice is to fully configure the database on the active node, and then bring the standby nodes online. Avoid joining the standby nodes to the cluster before importing large datasets or performing other data-intensive operations. Such operations cause extensive and continual ongoing updates to occur.

Note: In extreme cases, performance degradation from extensive ongoing updates may interrupt communications between a standby node and the active node. If a standby node loses contact with the active node for more than 10 seconds, it will go into discovery mode. At that point, however, it will find the active node, come out of discovery mode as a standby again, and then perform a remote restore.