• At most once—Messages may be lost but are never redelivered.
• Exactly once—this is what people actually want, each message is delivered once and only once.

It’s worth noting that this breaks down into two problems: the durability guarantees for publishing a message and the guarantees when consuming a message.

Many systems claim to provide “exactly once” delivery semantics, but it is important to read the fine print, most of these claims are misleading (i.e. they don’t translate to the case where consumers or producers can fail, cases where there are multiple consumer processes, or cases where data written to disk can be lost).

Prior to 0.11.0.0, if a producer failed to receive a response indicating that a message was committed, it had little choice but to resend the message. This provides at-least-once delivery semantics since the message may be written to the log again during resending if the original request had in fact succeeded. Since 0.11.0.0, the Kafka producer also supports an idempotent delivery option which guarantees that resending will not result in duplicate entries in the log. To achieve this, the broker assigns each producer an ID and deduplicates messages using a sequence number that is sent by the producer along with every message. Also beginning with 0.11.0.0, the producer supports the ability to send messages to multiple topic partitions using transaction-like semantics: i.e. either all messages are successfully written or none of them are. The main use case for this is exactly-once processing between Kafka topics (described below).

Not all use cases require such strong guarantees. For uses which are latency sensitive we allow the producer to specify the durability level it desires. If the producer specifies that it wants to wait on the message being committed this can take on the order of 10 ms. However the producer can also specify that it wants to perform the send completely asynchronously or that it wants to wait only until the leader (but not necessarily the followers) have the message.

1. It can read the messages, process the messages, and finally save its position. In this case there is a possibility that the consumer process crashes after processing messages but before saving its position. In this case when the new process takes over the first few messages it receives will already have been processed. This corresponds to the “at-least-once” semantics in the case of consumer failure. In many cases messages have a primary key and so the updates are idempotent (receiving the same message twice just overwrites a record with another copy of itself).

So what about exactly once semantics (i.e. the thing you actually want)? When consuming from a Kafka topic and producing to another topic (as in a application), we can leverage the new transactional producer capabilities in 0.11.0.0 that were mentioned above. The consumer’s position is stored as a message in a topic, so we can write the offset to Kafka in the same transaction as the output topics receiving the processed data. If the transaction is aborted, the consumer’s position will revert to its old value and the produced data on the output topics will not be visible to other consumers, depending on their “isolation level.” In the default “read_uncommitted” isolation level, all messages are visible to consumers even if they were part of an aborted transaction, but in “read_committed,” the consumer will only return messages from transactions which were committed (and any messages which were not part of a transaction).

When writing to an external system, the limitation is in the need to coordinate the consumer’s position with what is actually stored as output. The classic way of achieving this would be to introduce a two-phase commit between the storage of the consumer position and the storage of the consumers output. But this can be handled more simply and generally by letting the consumer store its offset in the same place as its output. This is better because many of the output systems a consumer might want to write to will not support a two-phase commit. As an example of this, consider a Kafka Connect connector which populates data in HDFS along with the offsets of the data it reads so that it is guaranteed that either data and offsets are both updated or neither is. We follow similar patterns for many other data systems which require these stronger semantics and for which the messages do not have a primary key to allow for deduplication.