Flow triggers
A trigger determines how a flow runs. The following are the supported flow trigger types:
-
Run on demand — Users manually run the flow as needed.
-
Run on event — Amazon AppFlow runs the flow in response to an event from an SaaS application.
-
Run on schedule — Amazon AppFlow runs the flow on a recurring schedule.
On demand flows
You can manually run on-demand flows as needed. You must run this type of flow each time you want to transfer the data. For more information, see Managing Amazon AppFlow flows.
Event-triggered flows
Amazon AppFlow runs event-triggered flows based on a specified change event in the source application.
This option is available only for SaaS applications that provide change events. You must choose the event when you choose the source.
Schedule-triggered flows
Amazon AppFlow runs schedule-triggered flows based on the schedule that you specify during flow setup. The scheduling frequency depends on the frequency supported by the source application.
You can choose either full or incremental data transfer for schedule-triggered flows.
Full transfer
When you select full transfer, Amazon AppFlow transfers a snapshot of all records at the time of the flow run from the source to the destination.
Incremental transfer
When you select incremental transfer, Amazon AppFlow transfers only the records that have been added or changed since the last successful flow run. You can also select a source timestamp field to specify how Amazon AppFlow identifies new or changed records. For example, if you have a Created Date timestamp field, choose this to instruct Amazon AppFlow to transfer only newly-created records (and not changed records) since the last successful flow run. The first schedule-triggered flow will pull 30 days of past records at the time of the first flow run.
Tip
To transfer records created or modified over a different time range other than the past 30 days at the time of the first flow run, set up the flow to be triggered on demand. You can then use the filter option to pull records over the desired time range. After the on-demand flow runs and pulls the initial set of records, edit the flow to be triggered on schedule so that subsequent flow runs transfer incremental data.
Offset option
Optionally, you can add a time offset (t) to the time range for the incremental transfer. The flow run will import records that were created or changed between the previous flow run and the specified offset prior to the current flow run. This feature can be used to accommodate any latencies in the source systems in timestamping changes to records. By choosing a sufficiently large offset, you can avoid missing records that changed in the source application close to the run time of the scheduled flow.
If a schedule-triggered flow runs at time instances T0, T1, T2, and so on, then records that are new or have changed between T0 minus t and T1 minus t will be imported from the source at T1, and those that have changed between T1 minus t and T2 minus t will be imported from the source at T2.
The total offset value can be longer than the schedule interval (for example, t can be longer than T1 minus T0), but it must be less than 10 hours. The default value is 0.
-
The flow run at T0 transfers records that changed between T0 minus 30 days and T0 minus t in the source application.
The flow run at T1 transfers records that changed between T0 minus t and T1 minus t in the source application.
The flow run at T2 transfers records that changed between T1 minus t and T2 minus t in the source application.
The flow run at T3 transfers records that changed between T2 minus t and T3 minus t in the source application.
