Kafka Integration

Kafka is a component in the Seldon Core 2 ecosystem, that provides scalable, reliable, and flexible communication for machine learning deployments. It serves as a strong backbone for building complex inference pipelines, managing high-throughput asynchronous predictions, and seamlessly integrating with event-driven systems—key features needed for contemporary enterprise-grade ML platforms.

An inference request is a request sent to a machine learning model to make a prediction or inference based on input data. It is a core concept in deploying machine learning models in production, where models serve predictions to users or systems in real-time or batch mode.

To explore this feature of Seldon Enterprise Platform, you need to integrate with Kafka. You can do so by either:

• integrating it through managed solution within your cloud provider or

• by deploying it directly within a Kubernetes cluster.

Note: Kafka is an external component outside the main Seldon stack. Therefore, it is the cluster administrator’s responsibility to administrate and manage the Kafka instance used by Seldon. For production installation it is highly recommended to use a managed Kafka instance.

For more information on how to set up a managed Kafka solution, refer to the following sections:

• Configuration examples provides the steps to configure some of the managed Kafka services.

• Securing Kafka provides more information about the encryption and authentication.