Configure service integration
Service integration lets you create a service and define its input parameters, response parameters, and the models to invoke. After you create and validate the service, you can lock the service and deploy it to a node to make it available for service invocation.
Prerequisites
The models required by the service are deployed.
Confirm the input and response parameters with the invoking system.
Step 1: Create a service
Before you can integrate a service, you must create it.
Log on to the Multi-party Secure Modeling console.
In the navigation pane on the left, click Model Deployment > Service Integration.
On the Service Integration page, click Create Service and configure the following parameters.
Parameter
Description
Version Orchestration Template
Select a blank template or a system-provided template as needed.
Service Name
Set the service name. The name must meet the following requirements:
Contain only Chinese characters, letters, digits, or underscores (_).
Be no more than 40 characters in length.
Unique Identifier
The unique identifier must consist of letters, digits, or underscores (_) and must be globally unique.
If a service identifier has been agreed upon offline, the unique identifier is the part of the service identifier that follows the last period (.). For example, if the service identifier is
dx.datalab.NODE_TEST_PREDICTION_bvyYhN7z.HB_MODEL, the unique identifier isHB_MODEL.Service Description (Optional)
The description of the service.
Version Name
The name of the service version. It can contain only digits, letters, and periods (.).
Version Description
The description of the version.
Click New to create the service.
Step 2: Orchestrate the flow
After the service is created, orchestrate its flow.
Log on to the Multi-party Secure Modeling console.
In the navigation pane on the left, click Model Deployment > Service Integration.
On the Service Integration page, click the service name, and then click the version number to open the flow orchestration page.
Orchestrate the flow.
Drag components from the toolbar on the left to the canvas on the right. Then, connect them with lines to build the service flow.
NoteAn orchestration flow must start with a Start component and terminate with an End component.
Ruleset: Supports conditional rules such as
CASE WHENstatements.Script: Supports adding Java code to process variables.
Model: Supports referencing models that are successfully deployed in Model Deployment.
Fork: Supports conditional rules such as
IF ELSEstatements.
Step 3: Validate the service
After you orchestrate the service flow, you can validate the service.
In the upper-right corner of the flow orchestration page for the service, click Validate > Single Validation.
On the Single Validation page, you can configure the service validation information.
Decision Input Parameters: Configure the input parameters for the decision.
Decision Response Parameters (Optional): Enter the expected values for comparison during validation.
Select Node to Initiate Invocation: The node that starts the service validation.
After completing the configuration, click Confirm.
Step 4: Submit the service
On the flow orchestration page for the service, click the Submit button in the upper-right corner.
In the confirmation dialog box, click Confirm.
You cannot orchestrate the flow for a submitted service. However, you can still validate it on the flow orchestration page.
You can delete a submitted service if it has not been deployed or if it has been unpublished after deployment.
Step 5: Deploy the service
Deploy the validated service to a node. After a successful deployment, you can invoke the service and put the model into production.
On the Service Integration page, click the name of a service, and then click Deploy next to the service version.
On the Service Deployment page, configure the information for the service deployment.
Parameter
Description
Service Name
Select the name of the service to deploy.
Version Name
Select the version of the service to deploy.
Deployment Node
Select the node that will initiate the invocation. For bidirectional invocation, select two nodes.
Click Submit Request to start the deployment.
Step 6: Adjust traffic
If multiple versions of the same service are deployed on a node, all versions share the same invocation interface. When the service is invoked, the system distributes traffic to different service versions based on the configured traffic splitting ratio. This allows for canary adjustments of service versions.
On the Service Integration page, click Deployment Details to the right of the service name.
On the service details page, click Adjust Traffic Splitting.
Traffic splitting policy: If multiple versions of a service are deployed on a node, the sum of the traffic ratios for all versions must be 100%.
When you are finished, click OK.