Dataflow effects

Effects describe which kinds of data operations a dataflow performs: reading, writing, or deleting data. Datahub calculates them from the nodes in the dataflow. They’re shown wherever a dataflow is exposed to other systems, for example as a capability in an MCP configuration.

Supported effects

Effect	Value	Description
Unknown	0	The effect could not be determined.
Read	1	Reads data in some way.
Write	2	Creates or updates data in some way.
Delete	4	Deletes data or marks data for deletion.

Effects are flags, so a dataflow can have several at once. The values add up:

Effects	Combined value	Meaning
Read + Write	3	Reads and creates or updates data.
Read + Delete	5	Reads and deletes data.
Write + Delete	6	Creates, updates, and deletes data.
Read + Write + Delete	7	Performs all three operation types.

Unknown has the value 0, so it can’t be combined with other effects.

Node effects

Each node in a dataflow contributes its own effects. Together they make up the effects of the dataflow.

Fixed effects

Most nodes have a predefined effect based on what they do:

Node category	Effect	Description
Import nodes	Read	Retrieve or read data from a source.
Export nodes	Write	Save or export data to a destination.

Configurable effects

Some nodes can perform different operations depending on how they’re configured. For these nodes, you set the effects yourself in the node properties:

• REST API nodes: a GET request reads data, while a POST or DELETE request changes or removes it.
• Database nodes, such as Microsoft SQL Server and MySQL: a query can read, change, or delete records.
• Execute Dataflow nodes (see below).

Make sure the configured effects match what the node actually does.

Execute Dataflow nodes

An Execute Dataflow node uses its own configured effects when the parent dataflow is evaluated. It does not inherit the calculated effects of the dataflow it calls.

For example:

• Dataflow A reads and updates information, so its calculated effects are Read + Write.
• Dataflow B contains an Execute Dataflow node that calls Dataflow A, configured with the effect Read.

When Dataflow B is evaluated, that node contributes only Read, even though Dataflow A does Read + Write.

Caution

Set the effects of an Execute Dataflow node to match what the called dataflow really does. Incorrect effects give external systems, such as MCP clients, misleading information about the dataflow.

When effects are calculated

Effects are calculated every time you save the dataflow:

1. The effects of all nodes are collected.
2. Configurable nodes contribute their configured effects.
3. The effects are combined into one value.
4. The result is stored as the dataflow’s calculated effects.

To change the effects of a dataflow, change the relevant nodes and save the dataflow again. Effects can’t be overridden elsewhere, for example in the MCP designer.

Tips

• Pay extra attention to REST API and database nodes, since they can perform several kinds of operations.
• Always check the configured effects of Execute Dataflow nodes.
• Save the dataflow after changing nodes or effect settings, so the effects are recalculated.
• Treat Unknown as a sign that the behavior couldn’t be determined: review the nodes and configure their effects where possible.