CheckpointFrequency
The CheckpointFrequency parameter determines how often the High-speed Apply Engine performs checkpoint processing to update the restart table with a statement ID that reflects the overall progress of all agents through the input file. When you restart an apply request, High-speed Apply Engine uses this information to speed up restart processing.
Attributes
This parameter has the following attributes:
Attribute | Value |
|---|---|
Section | [Restart] |
Abbreviation | None |
DBMS | Db2, Db2 LUW, Oracle |
Required? | No |
Valid values |
|
Default value | 5 |
Usage
Specify the number of commit actions that each agent performs before it performs checkpoint processing. Specify 0 to turn off checkpoint processing.
Use this parameter to balance the overhead of checkpoint processing against the benefits of shorter run time when you restart an apply request. Checkpoint processing imposes a small amount of overhead during an apply request. When you restart an apply request, however, previous checkpoint processing can provide a large benefit because High-speed Apply Engine can begin processing more quickly. (High-speed Apply Engine uses the statement ID in the restart table to read through the initial part of the input file without performing extra processing to determine if it has already processed an input statement.)
Consider the following situations:
- An apply request processes a large input file and failures occur frequently because of communication problems in your environment. In this case, set the CheckpointFrequency parameter to a lower value. Each agent performs checkpoint processing more frequently. The apply request can run slightly slower, but if it fails and you restart it, High-speed Apply Engine reads through the initial portion of the large input file quickly before it begins performing the extra processing. The restarted request runs faster.
- An apply request processes a smaller input file and you know from previous runs that failures are rare. In this case, set the CheckpointFrequency parameter to a higher value. Each agent performs checkpoint processing less frequently. The apply request runs slightly faster. If you have to restart the request, the restarted run takes longer.
Consider the following points as you set a value for CheckpointFrequency:
Your checkpoint frequency is a number of commit actions. With distribution by object, your commit actions are determined by the values of the StatementCount and Time parameters. (For more information, see StatementCount-parameter and Time.) Consider the values of these parameters as you set the checkpoint frequency. (With distribution by UR, the commit actions are determined by the original transactions in the input SQL statements.)
The number of commit actions is also influenced by the number of agents that process an apply request. Consider the values of your [Agent] parameters when you select a value for CheckpointFrequency.
- When High-speed Apply Engine processes data definition language (DDL) statements in an input file, it uses only one agent and issues a COMMIT statement after each DDL statement. This behavior can affect your checkpoint frequency.
Related topics