Working with Avro Serialized Data and Schema Registries
PreviousWorking with ProtoBuf Serialized Data and DeltaStream DescriptorsNextConfiguring Deserialization Error Handling
Last updated
Last updated
In streaming data stores such as and , producers send data events as bytes that must be interpreted by the data consumers. The most popular formats for data serialization include , , and . DeltaStream supports all of these.
This article focuses on . It shows how to create and use a to access that are necessary for data serialization/deserialization.
If you have a Data Store with entities whose data records are serialized with Avro, it's likely you also have a schema registry to manage the Avro schemas that are used to serialize and deserialize these data events. To do this in DeltaStream, you create a DeltaStream entity -- which is a wrapper around your schema registry -- and associate it with one or many data stores.
DeltaStream supports 2 types of schema registries (more are in development):
Confluent Cloud
Confluent Platform
To begin, create a schema registry in either the CLI or the UI SQL page. In the CLI, use the CREATE SCHEMA_REGISTRY command to create a DeltaStream :
In the above example, you're creating a CONFLUENT_CLOUD type schema registry named ConfluentCloudSR in AWS us-east-2. The uris value provided here is the URI from the Confluent Cloud dashboard, corresponding with the schema registry. Optionally, this schema registry can also have one of the following:
a key pair for credentials, which can be supplied with the confluent_cloud.key and confluent_cloud.secret properties
the properties.file property (see CREATE SCHEMA_REGISTRY for more details)
When you have defined schema registries, you can list them:
Next, associate that schema registry with the relevant data store.
You can describe a data store to determine whether it has a schema registry attached:
When you determine there is no schema registry attached to your data store, (kafka_store, above) you can use the UPDATE STORE DDL to point your data store to the schema registry.
When you attach a schema registry to a data store, then each time a DeltaStream command or query requires serializing or deserializing data from the data store’s entity, that data store uses the registry to look up schemas. When you work with Avro serialized data, DeltaStream requires that you attach the schema registry with the relevant Avro schemas to the data store containing that data.
Now you can successfully run commands such as PRINT ENTITY or write queries with DeltaStream objects that have Avro data formats. Below we have printed the pageviews_avro entity.
When you have set up a schema registry, you can easily read or write Avro-formatted data. The query below shows how to easily convert the JSON stream pageviews_json to a stream with an Avro key and value format called pageviews_converted_to_avro. See CREATE STREAM AS SELECT for more details.
When you create DeltaSTream objects using CREATE STREAM AS SELECT or CREATE CHANGELOG AS SELECT, and you specify a key or value format such as Avro for the sink object, DeltaStream automatically generates an Avro schema and adds it to the schema registry attached to the data store. In the example above, two schemas are generated for the entity pageviews_converted_to_avro -- one each for the key and the value. In this way these schemas are available if you ever need to consume from these entities outside of DeltaStream.
Finally, if the entity doesn’t have a descriptor, and the data store doesn’t have a schema registry (or it has a schema registry, but there is no corresponding schema in the registry), DeltaStream tries to deserialize the data in the entity as JSON.
In the case of the PRINT ENTITY command, if an entity in a data store has a , the descriptor is used for deserialization even if the data store has a schema registry. If the entity does not have a descriptor, the data store verifies that the schema registry contains a schema for the entity and then uses it for deserialization.