Improving target database performance with Fast Apply

Fast Apply is a product feature that provides opportunities to increase throughput and reduce latency when the apply process (your target database) is a performance bottleneck.

With Fast Apply you can choose from four different algorithms (or modes) that are appropriate for different business needs. Fast Apply is enabled with a subscription-level user exit in Management Console.

Resource footprint of CDC Replication

Using Fast Apply to improve performance will often result in an increased resource footprint of CDC Replication in the form of CPU and memory.

You should keep this in mind when deciding if this feature is appropriate for your environment. Fast Apply is probably not necessary or appropriate if your throughput and latency are adequate and meet your business needs. Similarly, Fast Apply may not be an appropriate choice if your target database is already constrained by limited resources and you do not intend to increase the available resources.

Restrictions when using Fast Apply

You should be aware of the following restrictions if you plan on using the Fast Apply feature in CDC Replication.

You cannot use Fast Apply if the following conditions exist:

  • You have tables with RI (Referential Integrity) constraints. If you choose to use RI constraints on your target tables, CDC Replication provides no guarantee that Fast Apply reordered operations will succeed.
  • You are replicating to a CDC Replication Engine for Netezza® Technology target.
  • You are using %GETCOL column functions in a derived expression on your target system.
  • You are using any user exits, user functions, or triggers that depend on the order that the operations are applied (for example if the user exit, function, or trigger looks at the values of other rows in that table or some other table).
  • You are using materialized views on your CDC Replication target system.
Note: If you decide to disable RI constraints for tables that are in replication scope, CDC Replication will replicate the data without any changes. There is no risk of data loss or data corruption.