Click here to view and discuss this page in DocCommentXchange. In the future, you will be sent there automatically.

SQL Anywhere 11.0.0 » SQL Remote » Replication Design for SQL Remote » Principles of SQL Remote design


Who gets what?

Each time a row in a table is inserted, deleted, or updated, a message has to be sent to those subscribed to the row. In addition, an update may cause the subscription expression to change, so that the statement is sent to some subscribers as a delete, some as an update, and some as an insert.

For more information about what statements get sent to which subscribers, see How statements are replicated.

For more information about subscriptions, see:

This section describes how SQL Remote sends the right operations to the right recipients.

The task of determining who gets what is divided between the database server and the Message Agent. The engine handles those aspects that are to do with publications, while the Message Agent handles aspects to do with subscriptions.

SQL Anywhere actions

SQL Anywhere evaluates the subscription expression for each update made to a table that is part of a publication. It adds the value of the expression to the log, both before and after the update.

For a table that is part of more than one publication, the subscription expression is evaluated before and after the update for each publication.

The addition of information to the log can affect performance in the following cases:

  • Expensive expressions   When a subscription expression is expensive to evaluate, it can affect performance.

  • Many publications   When a table belongs to many publications, many expressions must be evaluated. In contrast, the number of subscriptions is irrelevant.

  • Many-valued expressions   Some expressions are many-valued. This can lead to much additional in formation in the transaction log, with a corresponding effect on performance.

Message Agent actions

The Message Agent reads the evaluated subscription expressions or subscription column entries from the transaction log, and matches the before and after values against the subscription value for each subscriber to the publication. In this way, the Message Agent can send the correct operations to each subscriber.

While large numbers of subscribers do not have any impact on server performance, they can impact Message Agent performance. Both the work in matching subscription values against large numbers of subscription values, and the work in sending the messages, can be demanding.