aromanenko-dev opened a new issue, #26262:
URL: https://github.com/apache/beam/issues/26262

   ### What needs to happen?
   
   From [email 
thread](https://lists.apache.org/thread/97dkw2rc4tf1f0gfmnfgdfgdg7nvbwmw):
   
   "I am trying to understand the effect of schema registry on our pipeline's 
performance. In order to do sowe created a very simple pipeline that reads from 
kafka, runs a simple transformation of adding new field and writes of kafka.  
the messages are in avro format
   
   I ran this pipeline with 3 different options on same configuration : 1 kafka 
partition, 1 task manager, 1 slot, 1 parallelism:
   
   * when i used apicurio as the schema registry i was able to process only 
2000 messages per second
   * when i used confluent schema registry i was able to process 7000 messages 
per second
   * when I did not use any schema registry and used plain avro 
deserializer/serializer i was able to process 30K messages per second.
   
   ```
   KafkaIO.<String, T>read()
           .withBootstrapServers(bootstrapServers)
           .withTopic(topic)
           .withConsumerConfigUpdates(Map.ofEntries(
                   Map.entry("schema.registry.url", registryURL),
                   Map.entry(ConsumerConfig.GROUP_ID_CONFIG, consumerGroup+ 
UUID.randomUUID()),
           ))
           .withKeyDeserializer(StringDeserializer.class)
           .withValueDeserializerAndCoder((Class) 
io.confluent.kafka.serializers.KafkaAvroDeserializer.class, 
AvroCoder.of(avroClass));
   ```
   
    
   
   I have made the suggested change and used 
`ConfluentSchemaRegistryDeserializerProvider`
   the results are slightly  better.. average of 8000 msg/sec "
   
   ### Issue Priority
   
   Priority: 2 (default / most normal work should be filed as P2)
   
   ### Issue Components
   
   - [ ] Component: Python SDK
   - [X] Component: Java SDK
   - [ ] Component: Go SDK
   - [ ] Component: Typescript SDK
   - [ ] Component: IO connector
   - [ ] Component: Beam examples
   - [ ] Component: Beam playground
   - [ ] Component: Beam katas
   - [ ] Component: Website
   - [ ] Component: Spark Runner
   - [ ] Component: Flink Runner
   - [ ] Component: Samza Runner
   - [ ] Component: Twister2 Runner
   - [ ] Component: Hazelcast Jet Runner
   - [ ] Component: Google Cloud Dataflow Runner


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to