Setting up server-initiated synchronization in a synchronization model

Server-initiated synchronization allows you to initiate synchronization on the client database when something changes on the consolidated database. You can enable server-initiated synchronization in a synchronization model. This method allows you to deploy a limited version of server-initiated synchronization that is easy to set up and run.

When enabling server-initiated synchronization in a synchronization model, the MobiLink server uses a download_cursor script for a table to determine when to initiate a synchronization. It uses the download_cursor script to generate a request_cursor for the Notifier. You cannot customize your request_cursor using this method.

 To set up server-initiated synchronization in a synchronization model
  1. In the left pane of Sybase Central, expand MobiLink 12, your MobiLink project name, Synchronization Models, and then select your synchronization model name.

  2. Open the Notification tab in the right pane.

  3. Select Enable Server-Initiated Synchronization.

  4. Select a consolidated database table to use for notification.

    A change to the data in this table results in a notification being sent to the remote database. The notification triggers a synchronization.

    Note

    You can only choose tables for this purpose for which you have defined a timestamp-based download cursor (the default). The notification is based on the contents of the download cursor.

  5. Select a polling interval.

    The polling interval is the time between polls. You can choose a predefined polling interval or you can specify your own. The default is 30 seconds.

    If the Notifier loses the database connection, it recovers automatically at the first polling interval after the database becomes available again.

  6. Change the isolation level of the Notifier's database connection. (Optional)

    The default is read committed. Be aware of the consequences of setting the isolation level. Higher levels increase contention, and may adversely affect performance. Isolation level 0 allows reads of uncommitted data—data which may eventually be rolled back.

 See also