Executing Business Logic During Merge Synchronization
The business logic handler framework allows you to write a managed code assembly that is called during the merge synchronization process. The assembly includes business logic that can respond to a number of conditions during synchronization: data changes, conflicts, and errors. The business logic handler framework provides a simple programming model, and the data that the merge process provides to your assembly is in the form of an ADO.NET data set, so you can leverage knowledge of ADO.NET rather than learning a proprietary interface. For more information on programming business logic handlers, see:
The application programming interface (API) reference: Microsoft.SqlServer.Replication.BusinessLogicSupport
Instructions on how to implement a business logic handler: How to: Implement a Business Logic Handler for a Merge Article (Replication Programming)
Uses for Business Logic Handlers
The merge synchronization process can invoke business logic handlers to perform:
Custom change handling
Custom conflict resolution
Custom error resolution
The business logic handler you specify is executed for every row that is synchronized. Complex logic and calls to other applications or network services can impact performance.
Custom Change Handling
The business logic handler can be invoked during the processing of non-conflicting data changes and can perform one of three actions:
Reject the data
This is useful for applications that do not want changes propagated to or from a given Subscriber. For example, an administrator could filter out inserts that do not belong in the Subscriber’s partition, or possibly reject deletes performed at a Subscriber. As another example, an application could reject an order entered at a Subscriber because the inventory is no longer available.
Accept the data
This is useful for applications in which it is necessary to review data changes made at either the Publisher or Subscriber before allowing them to be propagated. For example, a mid-tier application could examine new orders coming in from the field and integrate with a procurement workflow process in the mid-tier.
Apply custom data
This is useful for applications that need to override specific data values or operations. For example, an application could transform a row delete into a special update that sets a status column in the row to a value of "deleted" and then tracks the identity of the client performing the delete. This might be useful for auditing or workflow purposes.
Custom Conflict Resolution
Merge replication provides conflict detection and resolution, allowing you to accept a default resolution strategy or choose custom resolution for conflicts. For more information, see Advanced Merge Replication Conflict Detection and Resolution. The business logic handler can be invoked during the processing of conflicting data changes and can perform one of two actions:
Accept default resolution
This is useful for applications that might need to review the conflict, perform additional actions, and possibly log a custom conflict log message.
Perform custom resolution
This is useful for applications that might need to select data values that are specific to their business logic and supply the synchronization process with this custom dataset. For example, an application could provide a new version of the winning row by combining values from the Publisher and Subscriber data sets.
Custom Error Resolution
Custom logic can be invoked during the propagation of changes that result in errors. The logic can perform one of two actions:
Accept default error resolution
This is useful for applications that might need to review the error and perform additional action and possibly log a custom error log message.
Accept custom error resolution
This is useful for applications that might need to select data values that are specific to their business logic and supply the synchronization process with this custom dataset. For example, if the replication process encounters a duplicate key violation, the business logic handler could provide a new version of the data change in which the key will no longer conflict. Changes made at the Publisher and Subscriber can then persist in the database, and the replication process doesn't have to compensate for the failed insert with a delete.
Deployment Scenarios for Business Logic Handlers
Business logic handlers can be deployed at:
The Distributor. Use a push subscription so that business logic is executed at the Distributor.
The Subscriber. Use a pull subscription so that business logic is executed at the Subscriber.
An Internet Information Services (IIS) server if Web synchronization is used. Use a pull subscription synchronized with Web synchronization, and the business logic handler will execute at the IIS Server.