Using Streaming with Apache Kafka

Oracle Cloud Infrastructure Streaming lets users of Apache Kafka offload the setup, maintenance, and infrastructure management that hosting your own Zookeeper and Kafka cluster requires.

Streaming is compatible with most Kafka APIs, allowing you to use applications written for Kafka to send messages to and receive messages from the Streaming service without having to rewrite your code. See Using Kafka APIs for more information.

Streaming can also utilize the Kafka Connect ecosystem to interface directly with external sources like databases, object stores, or any microservice on the Oracle Cloud. Kafka connectors can easily and automatically create, publish to, and deliver topics while taking advantage of the Streaming service's high throughput and durability. See Using Kafka Connect for more information.

Use cases for Streaming and Kafka include:

  • Move data from Streaming to Autonomous Data Warehouse via the JDBC Connector to perform advanced analytics and visualization.

  • Use the Oracle GoldenGate connector for Big Data to build an event-driven application.

  • Move data from Streaming to Oracle Object Storage via the HDFS/S3 Connector for long term storage, or to run Hadoop/Spark jobs.

Kafka Clients

While many Kafka clients are available, we recommend the clients that have been fully tested and certified to work with the Streaming service.

Streaming supports all versions of apache-kafka-java.

Streaming also supports the following Kafka clients on a best-effort basis:

Requirements and Limitations

The implementation of Streaming's Kafka compatibility results in the following configurations, limitations, and behaviors.

Lossless Configuration

Streaming only supports lossless Kafka configurations. Data is replicated three ways. Messages from producers do not initiate an acknowledgment (ACK) from Streaming until at least two replicas are in sync.

Unique Stream Names

If you have streams with the same names in a compartment, you can't use Kafka with Streaming until you delete the duplicated streams, unless the streams are in different stream pools. Two streams with the same name can exist in the same compartment only if the streams are in different stream pools.

Duplicate stream names otherwise manifest through an "authentication failed" error. If you do not want to delete your streams, contact the Streaming team so we can rename your streams without data loss.

Load Balancing Connection Recycling

Because the Kafka protocol uses long-lived TCP connections, the Streaming Kafka compatibility layer implements a load balancing mechanism to periodically balance connections between front-end nodes. This mechanism periodically closes connections to force new ones. Most Kafka SDKs handle these disconnections automatically when consuming, but producing to Streaming using the Kafka APIĀ might raise disconnection errors. Disconnections can be mitigated by adding retries to your requests. Retries are part of the Kafka SDK and are automatically enabled, and you can explicitly configure their behavior.