Chapter 7. Validating schemas using Kafka client serializers/deserializers in Java

Service Registry decouples schema management from client application configuration. You can enable a Java client application to use a schema from Service Registry by specifying its URL in your client code.

You can store the schemas in the registry to serialize and deserialize messages, which are referenced from your client applications to ensure that the messages that they send and receive are compatible with those schemas. Kafka client applications can push or pull their schemas from Service Registry at runtime.

Schemas can evolve, so you can define rules in Service Registry, for example, to ensure that schema changes are valid and do not break previous versions used by applications. Service Registry checks for compatibility by comparing a modified schema with previous schema versions.

Service Registry schema technologies

Service Registry provides schema registry support for schema technologies such as:

These schema technologies can be used by client applications through the Kafka client serializer/deserializer (SerDe) services provided by Service Registry. The maturity and usage of the SerDe classes provided by Service Registry might vary. The sections that follow provide more details about each schema type.

Producer schema configuration

A producer client application uses a serializer to put the messages that it sends to a specific broker topic into the correct data format.

To enable a producer to use Service Registry for serialization:

After registering your schema, when you start Kafka and Service Registry, you can access the schema to format messages sent to the Kafka broker topic by the producer. Alternatively, depending on configuration, the producer can automatically register the schema on first use.

If a schema already exists, you can create a new version using the registry REST API based on compatibility rules defined in Service Registry. Versions are used for compatibility checking as a schema evolves. A group ID, artifact ID, and version represents a unique tuple that identifies a schema.

Consumer schema configuration

A consumer client application uses a deserializer to get the messages that it consumes from a specific broker topic into the correct data format.

To enable a consumer to use Service Registry for deserialization:

When locating the global ID in the message payload, the format of the data begins with a magic byte, used as a signal to consumers, followed by the global ID, and the message data as normal. For example:

# ...
[MAGIC_BYTE]
[GLOBAL_ID]
[MESSAGE DATA]

Then when you start Kafka and Service Registry, you can access the schema to format messages received from the Kafka broker topic.

The content ID does not uniquely identify a version, but uniquely identifies the version content only. If multiple versions share the exact same content, they have a different global ID but the same content ID. The Confluent schema registry uses content ID by default.

The Kafka client serializer uses lookup strategies to determine the artifact ID and global ID under which the message schema is registered in Service Registry. For a given topic and message, you can use different implementations of the ArtifactResolverStrategy Java interface to return a reference to an artifact in the registry.

The classes for each strategy are in the io.apicurio.registry.serde.strategy package. Specific strategy classes for Avro SerDe are in the io.apicurio.registry.serde.avro.strategy package. The default strategy is the TopicIdStrategy, which looks for Service Registry artifacts with the same name as the Kafka topic receiving messages.

public ArtifactReference artifactReference(String topic, boolean isKey, T schema) {
        return ArtifactReference.builder()
                .groupId(null)
                .artifactId(String.format("%s-%s", topic, isKey ? "key" : "value"))
                .build();

Which lookup strategy you use depends on how and where you store your schema. For example, you might use a strategy that uses a record ID if you have different Kafka topics with the same Avro message type.

ArtifaceResolverStrategy interface

The artifact resolver strategy provides a way to map the Kafka topic and message information to an artifact in Service Registry. The common convention for the mapping is to combine the Kafka topic name with the key or value, depending on whether the serializer is used for the Kafka message key or value.

However, you can use alternative conventions for the mapping by using a strategy provided by Service Registry, or by creating a custom Java class that implements io.apicurio.registry.serde.strategy.ArtifactResolverStrategy.

Strategies to return an artifact reference

Service Registry provides the following strategies to return an artifact reference based on an implementation of ArtifaceResolverStrategy:

DefaultSchemaResolver interface

The default schema resolver locates and identifies the specific version of the schema registered under the artifact reference provided by the artifact resolver strategy. Every version of every artifact has a single globally unique identifier that can be used to retrieve the content of that artifact. This global ID is included in every Kafka message so that a deserializer can properly fetch the schema from Apicurio Registry.

The default schema resolver can look up an existing artifact version, or it can register one if not found, depending on which strategy is used. You can also provide your own strategy by creating a custom Java class that implements io.apicurio.registry.serde.SchemaResolver. However, it is recommended to use the DefaultSchemaResolver and specify configuration properties instead.

Strategies to return a global ID

When using the DefaultSchemaResolver, you can configure its behavior using application properties. The following table shows some commonly used examples:

You can configure specific client serializer/deserializer (SerDe) services and schema lookup strategies directly in a client application using the example constants shown in this section. Alternatively, you can configure the corresponding Service Registry application properties in a file or an instance.

The following sections show examples of SerDe constants and configuration options.

Configuration for SerDe services

public class SerdeConfig {

   public static final String REGISTRY_URL = "apicurio.registry.url"; 1
   public static final String ID_HANDLER = "apicurio.registry.id-handler"; 2
   public static final String ENABLE_CONFLUENT_ID_HANDLER = "apicurio.registry.as-confluent"; 3

Configuration for SerDe lookup strategies

public class SerdeConfig {

   public static final String ARTIFACT_RESOLVER_STRATEGY = "apicurio.registry.artifact-resolver-strategy";
...

Configuration for Kafka converters

public class SerdeBasedConverter<S, T> extends SchemaResolverConfigurer<S, T> implements Converter, Closeable {

   public static final String REGISTRY_CONVERTER_SERIALIZER_PARAM = "apicurio.registry.converter.serializer"; 1
   public static final String REGISTRY_CONVERTER_DESERIALIZER_PARAM = "apicurio.registry.converter.deserializer"; 2

Configuration for different schema types

For details on how to configure SerDe for different schema technologies, see the following:

When using schemas in your Kafka client applications, you must choose which specific schema type to use, depending on your use case. Service Registry provides SerDe Java classes for Apache Avro, JSON Schema, and Google Protobuf. The following sections explain how to configure Kafka applications to use each type.

You can also use Kafka to implement custom serializer and deserializer classes, and leverage Service Registry functionality using the Service Registry REST Java client.

Kafka application configuration for serializers/deserializers

Using the SerDe classes provided by Service Registry in your Kafka application involves setting the correct configuration properties. The following simple Avro examples show how to configure a serializer in a Kafka producer application and how to configure a deserializer in a Kafka consumer application.

// Create the Kafka producer
private static Producer<Object, Object> createKafkaProducer() {
    Properties props = new Properties();

    // Configure standard Kafka settings
    props.putIfAbsent(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, SERVERS);
    props.putIfAbsent(ProducerConfig.CLIENT_ID_CONFIG, "Producer-" + TOPIC_NAME);
    props.putIfAbsent(ProducerConfig.ACKS_CONFIG, "all");

    // Use Service Registry-provided Kafka serializer for Avro
    props.putIfAbsent(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, StringSerializer.class.getName());
    props.putIfAbsent(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, AvroKafkaSerializer.class.getName());

    // Configure the Service Registry location
    props.putIfAbsent(SerdeConfig.REGISTRY_URL, REGISTRY_URL);

    // Register the schema artifact if not found in the registry.
    props.putIfAbsent(SerdeConfig.AUTO_REGISTER_ARTIFACT, Boolean.TRUE);

    // Create the Kafka producer
    Producer<Object, Object> producer = new KafkaProducer<>(props);
    return producer;
}

// Create the Kafka consumer
private static KafkaConsumer<Long, GenericRecord> createKafkaConsumer() {
    Properties props = new Properties();

    // Configure standard Kafka settings
    props.putIfAbsent(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, SERVERS);
    props.putIfAbsent(ConsumerConfig.GROUP_ID_CONFIG, "Consumer-" + TOPIC_NAME);
    props.putIfAbsent(ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG, "true");
    props.putIfAbsent(ConsumerConfig.AUTO_COMMIT_INTERVAL_MS_CONFIG, "1000");
    props.putIfAbsent(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, "earliest");

    // Use Service Registry-provided Kafka deserializer for Avro
    props.putIfAbsent(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class.getName());
    props.putIfAbsent(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, AvroKafkaDeserializer.class.getName());

    // Configure the Service Registry location
    props.putIfAbsent(SerdeConfig.REGISTRY_URL, REGISTRY_URL);

    // No other configuration needed because the schema globalId the deserializer uses is sent
    // in the payload. The deserializer extracts the globalId and uses it to look up the schema
    // from the registry.

    // Create the Kafka consumer
    KafkaConsumer<Long, GenericRecord> consumer = new KafkaConsumer<>(props);
    return consumer;
}

props.putIfAbsent(SerdeConfig.ENABLE_HEADERS, "true")

props.putIfAbsent(SerdeConfig.VALIDATION_ENABLED, Boolean.FALSE)

After you have defined a schema in the appropriate format, such as Apache Avro, you can add the schema to Service Registry.

You can add the schema using the following approaches:

Client applications cannot use Service Registry until you have registered your schemas.

Service Registry web console

When Service Registry is installed, you can connect to the web console from the ui endpoint:

http://MY-REGISTRY-URL/ui

From the console, you can add, view and configure schemas. You can also create the rules that prevent invalid content being added to the registry.

Curl command example

 curl -X POST -H "Content-type: application/json; artifactType=AVRO" \
   -H "X-Registry-ArtifactId: share-price" \ 1
   --data '{
     "type":"record",
     "name":"price",
     "namespace":"com.example",
     "fields":[{"name":"symbol","type":"string"},
     {"name":"price","type":"string"}]}'
   https://my-cluster-my-registry-my-project.example.com/apis/registry/v2/groups/my-group/artifacts -s 2

Maven plug-in example

<plugin>
  <groupId>io.apicurio</groupId>
  <artifactId>apicurio-registry-maven-plugin</artifactId>
  <version>${apicurio.version}</version>
  <executions>
      <execution>
        <phase>generate-sources</phase>
        <goals>
            <goal>register</goal>  1
        </goals>
        <configuration>
            <registryUrl>http://REGISTRY-URL/apis/registry/v2</registryUrl> 2
            <artifacts>
                <artifact>
                    <groupId>TestGroup</groupId> 3
                    <artifactId>FullNameRecord</artifactId>
                    <file>${project.basedir}/src/main/resources/schemas/record.avsc</file>
                    <ifExists>FAIL</ifExists>
                </artifact>
                <artifact>
                    <groupId>TestGroup</groupId>
                    <artifactId>ExampleAPI</artifactId> 4
                    <type>GRAPHQL</type>
                    <file>${project.basedir}/src/main/resources/apis/example.graphql</file>
                    <ifExists>RETURN_OR_UPDATE</ifExists>
                    <canonicalize>true</canonicalize>
                </artifact>
            </artifacts>
        </configuration>
    </execution>
  </executions>
 </plugin>

Configuration using a producer client example

String registryUrl_node1 = PropertiesUtil.property(clientProperties, "registry.url.node1",
    "https://my-cluster-service-registry-myproject.example.com/apis/registry/v2"); 1
try (RegistryService service = RegistryClient.create(registryUrl_node1)) {
    String artifactId = ApplicationImpl.INPUT_TOPIC + "-value";
    try {
        service.getArtifactMetaData(artifactId); 2
    } catch (WebApplicationException e) {
        CompletionStage <ArtifactMetaData> csa = service.createArtifact(
            ArtifactType.AVRO,
            artifactId,
            new ByteArrayInputStream(LogInput.SCHEMA$.toString().getBytes())
        );
        csa.toCompletableFuture().get();
    }
}

This procedure describes how to configure a Kafka consumer client written in Java to use a schema from Service Registry.

This procedure describes how to configure a Kafka producer client written in Java to use a schema from Service Registry.

This procedure describes how to configure a Kafka Streams client written in Java to use an Apache Avro schema from Service Registry.