Chapter 5. Kafka producer configuration tuning

Connection and serializer properties are required for every producer. Generally, it is good practice to add a client id for tracking, and use compression on the producer to reduce batch sizes in requests.

In a basic producer configuration:

# ...
bootstrap.servers=localhost:9092 1
key.serializer=org.apache.kafka.common.serialization.StringSerializer 2
value.serializer=org.apache.kafka.common.serialization.StringSerializer 3
client.id=my-client 4
compression.type=gzip 5
# ...

Message delivery acknowledgments minimize the likelihood that messages are lost. Acknowledgments are enabled by default with the acks property set at acks=all.

# ...
acks=all 1
# ...

The acks=all setting offers the strongest guarantee of delivery, but it will increase the latency between the producer sending a message and receiving acknowledgment. If you don’t require such strong guarantees, a setting of acks=0 or acks=1 provides either no delivery guarantees or only acknowledgment that the leader replica has written the record to its log.

With acks=all, the leader waits for all in-sync replicas to acknowledge message delivery. A topic’s min.insync.replicas configuration sets the minimum required number of in-sync replica acknowledgements. The number of acknowledgements include that of the leader and followers.

A typical starting point is to use the following configuration:

When you create a topic, you can override the default replication factor. You can also override min.insync.replicas at the topic level in the topic configuration.

AMQ Streams uses this configuration in the example configuration files for multi-node deployment of Kafka.

The following table describes how this configuration operates depending on the availability of followers that replicate the leader replica.

A topic replication factor of 3 creates one leader replica and two followers. In this configuration, the producer can continue if a single follower is unavailable. Some delay can occur whilst removing a failed broker from the in-sync replicas or a creating a new leader. If the second follower is also unavailable, message delivery will not be successful. Instead of acknowledging successful message delivery, the leader sends an error (not enough replicas) to the producer. The producer raises an equivalent exception. With retries configuration, the producer can resend the failed message request.

Idempotent producers avoid duplicates as messages are delivered exactly once. IDs and sequence numbers are assigned to messages to ensure the order of delivery, even in the event of failure. If you are using acks=all for data consistency, using idempotency makes sense for ordered delivery. Idempotency is enabled for producers by default. With idempotency enabled, you can set the number of concurrent in-flight requests to a maximum of 5 for message ordering to be preserved.

# ...
enable.idempotence=true 1
max.in.flight.requests.per.connection=5 2
acks=all 3
retries=2147483647 4
# ...

If you choose not to use acks=all and disable idempotency because of the performance cost, set the number of in-flight (unacknowledged) requests to 1 to preserve ordering. Otherwise, a situation is possible where Message-A fails only to succeed after Message-B was already written to the broker.

# ...
enable.idempotence=false 1
max.in.flight.requests.per.connection=1 2
retries=2147483647
# ...

Idempotence is useful for exactly once writes to a single partition. Transactions, when used with idempotence, allow exactly once writes across multiple partitions.

Transactions guarantee that messages using the same transactional ID are produced once, and either all are successfully written to the respective logs or none of them are.

# ...
enable.idempotence=true
max.in.flight.requests.per.connection=5
acks=all
retries=2147483647
transactional.id=UNIQUE-ID 1
transaction.timeout.ms=900000 2
# ...

The choice of transactional.id is important in order that the transactional guarantee is maintained. Each transactional id should be used for a unique set of topic partitions. For example, this can be achieved using an external mapping of topic partition names to transactional ids, or by computing the transactional id from the topic partition names using a function that avoids collisions.

Usually, the requirement of a system is to satisfy a particular throughput target for a proportion of messages within a given latency. For example, targeting 500,000 messages per second with 95% of messages being acknowledged within 2 seconds.

It’s likely that the messaging semantics (message ordering and durability) of your producer are defined by the requirements for your application. For instance, it’s possible that you don’t have the option of using acks=0 or acks=1 without breaking some important property or guarantee provided by your application.

Broker restarts have a significant impact on high percentile statistics. For example, over a long period the 99th percentile latency is dominated by behavior around broker restarts. This is worth considering when designing benchmarks or comparing performance numbers from benchmarking with performance numbers seen in production.

Depending on your objective, Kafka offers a number of configuration parameters and techniques for tuning producer performance for throughput and latency.

At which point the send() method returns. So the time send() is blocked is determined by:

Batches will remain in the queue until one of the following occurs:

Look at the configuration for batching and buffering to mitigate the impact of send() blocking on latency.

# ...
linger.ms=100 1
batch.size=16384 2
buffer.memory=33554432 3
# ...

You can also direct messages to a specified partition by writing a custom partitioner to replace the default.

# ...
delivery.timeout.ms=120000 1
partitioner.class=my-custom-partitioner 2

# ...