Perform Translation Through AWS Sagemaker

Before implementing LiveRamp's AWS Embedded Transcoder, make sure that the following prerequisites have been met:

To deploy LiveRamp's AWS Embedded Transcoder and perform translation, perform these overall steps:

See the sections below for detailed information on performing these steps.

The LiveRamp Embedded Transcoder is deployed by running a template in the form of a YAML file within AWS CloudFormation. The file is contained inside a ZIP archive that you can get from a public GitHub repository.

The template installs a proxy service (written as an AWS Lambda function) and a SageMaker instance inside a Virtual Private Cloud. The VPC settings place the Transcoder deployment into network isolation mode, which only allows incoming requests through the proxy server to be processed.

When the wizard closes you are returned to the Stacks section and the new deployment will be listed as the latest event, along with its status as CREATE_IN_PROGRESS. The installation time is variable, depending on several factors such as the data center used, network traffic, the template that you chose, and more. A typical deployment installation will take about 7 to 15 minutes.

When the deployment is complete, all of the resources will show a CREATE_COMPLETE status. The Endpoint resource is the last resource to be completed.

Before making calls, you can use the AWS API Gateway to determine the appropriate URL to use for each transcoder endpoint (/init and /request).

Before performing translation, you need to initialize the transcoder and receive the required session key used to make translation calls. The session key is in the form of a "smart token" that contains the permissions for each source domain and each target domain that you are allowed to use in translation operations.

To initialize the transcoder and receive the required session key, call the /init endpoint with an rpc value of "init".

The following parameters are required (provided in the credentials previously sent by LiveRamp):

The cURL request to initialize the transcoder follows the format of the example below:

curl --location --request POST 'https://<hostname>.execute-api.<region>.amazonaws.com/Prod/auth/init' \
--header 'Content-Type: application/json' \
--data-raw '{
    "httpMethod": "POST",
    "client_id": "<client id>",
    "client_secret": "<client secret>",
“rpc": "init"

}'

The call returns the following response:

{
"transcoder_response": "Transcoder Initialization successful",
"smart_token": "xxxxxxxxxxxx"
}

A valid session key is required to perform translation. The session key contained within the smart token that is returned from the call to the /init endpoint expires after 15 minutes.

To refresh the session key, call the /request endpoint with an rpc value of "refresh".

The following parameters are required:

The cURL request to get a session key follows the format of the example below:

curl --location --request POST 'https://<hostname>.execute-api.<region>.amazonaws.com/Prod/auth/request' \
--header 'Content-Type: application/json' \
--data-raw '{
    "httpMethod": "POST",
    "client_id": "<client id>",
    "client_secret":"<client secret>",
    "rpc" : "refresh"
}'

The call returns the following response:

{
"smart_token": "xxxxxxxxxxxx"
}

To perform RampID translation, make a call to the /request endpoint with an rpc value of "request".

The following parameters are required:

Each JSON object must contain the following parameters:

Within one call, you can translate into multiple domains (if you're authorized to do so) by using different target domain encoding values.

The cURL call to perform translation follows the format of the example below:

curl --location --request POST https://<hostname>.execute-api.<region>.amazonaws.com/Pprod/auth/request' \
--header 'Content-Type: application/json' \
--data-raw '{
    "httpMethod": "POST",
    "transcode_data" : [{
            	"value": “<RampID value>”,
            "targetDomain": “<target domain encoding>”,
            "type": "rampidtrampid"
        },
        {
            "value": “<RampID value>”,
            "targetDomain": “<target domain encoding>”,
            "type": "rampidtrampid"
        }
],
    "smart_token" : "<smart token>",
    "rpc" : "request"
}'

The call returns the following response:

If any validations fail, an error response is returned. Common errors include:

To minimize costs when the work is completed, a user with privileges to delete stacks can delete the Transcoder deployment:

There are multiple logs and metrics captured throughout the system. Some logs are stored only in the account where the appliance is deployed and if an issue arises, it is at the discretion of the AWS Account Owner on the logs and metrics to communicate to LiveRamp.

Other metrics are sent to LiveRamp automatically, including usage metrics (number of successful translation calls by domain, to assist with billing) and authorization metrics (token validation failed), for example.

Other metrics may be stored in AWS CloudWatch on the Account of the deployed appliance. If troubleshooting an issue with LiveRamp, LiveRamp may request copies, screenshots, or dumps of logs and metrics in order to better assist with troubleshooting.

The Embedded Transcoder has two API endpoints: /init and /request. They are documented in the following two tables.

curl --location --request POST 'https://<hostname>.execute-api.<region>.amazonaws.com/Prod/auth/init' \
--header 'Content-Type: application/json' \
--data-raw '{
    "httpMethod": "POST",
    "client_id": "<client id>",
    "client_secret": "<client secret>",
“rpc": "init"

}'

{
"transcoder_response": "Transcoder Initialization successful",
"smart_token": "xxxxxxxxxxxx"
}

curl --location --request POST https://<hostname>.execute-api.<region>.amazonaws.com/Pprod/auth/request' \
--header 'Content-Type: application/json' \
--data-raw '{
    "httpMethod": "POST",
    "transcode_data" : [{
            	"value": “<RampID value>”,
            "targetDomain": “<target domain encoding>”,
            "type": "rampidtrampid"
        },
        {
            "value": “<RampID value>”,
            "targetDomain": “<target domain encoding>”,
            "type": "rampidtrampid"
        }
],
    "smart_token" : "<smart token>",
    "rpc" : "request"
}'

curl --location --request POST 'https://<hostname>.execute-api.<region>.amazonaws.com/Prod/auth/request' \
--header 'Content-Type: application/json' \
--data-raw '{
    "httpMethod": "POST",
    "client_id": "<client id>",
    "client_secret":"<client secret>",
    "rpc" : "refresh"
}'

{
"smart_token": "xxxxxxxxxxxx"
}

When working with the API you may sometimes receive an error code in the response. Refer to the following table for a description of errors you may encounter.