Kafka guarantees message order only within an individual partition—not across an entire multi-partition topic. In Go, keep related events ordered by giving them a stable key and configuring a producer balancer that sends that key to the same partition. Then make sure consumer workers do not commit past unfinished work if processing order matters.
What ordering Kafka guarantees
A Kafka topic is made up of partitions, and each partition is an ordered log. Kafka’s documentation states: “Messages sent by a producer to a particular topic partition will be appended in the order they are sent.” A consumer reading that partition sees records in the order stored in its log. Apache Kafka documentation
There is no Kafka-defined total order across multiple partitions. If two records are in different partitions, their relative arrival or processing time does not establish which one is first for the topic as a whole.
Keep related events in the same partition
Producers determine which partition receives a record. A common strategy is to use a key that represents the entity whose events must stay in sequence: for example, an account ID for balance changes. With a compatible key-based partitioner, records with the same key map to the same partition, where Kafka preserves their order. Kafka producer configuration
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match#1 Best Overall
This gives you per-key ordering, not a global sequence for all keys. It also depends on using a stable key and consistent partitioning behavior: if related records go to different partitions, Kafka cannot order them relative to one another.
Configure the balancer in kafka-go
If you use github.com/segmentio/kafka-go, inspect the Writer configuration and set its Balancer explicitly. The library documents Hash for routing records by key; other choices, such as round-robin and least-bytes, distribute records differently and do not provide the same per-key routing guarantee. Check the documentation for the version in your dependency rather than assuming a universal Go-client default. kafka-go balancers
writer := &kafka.Writer{
Addr: kafka.TCP("localhost:9092"),
Topic: "account-events",
Balancer: &kafka.Hash{},
}
err := writer.WriteMessages(ctx,
kafka.Message{Key: []byte("account-42"), Value: []byte("debit")},
kafka.Message{Key: []byte("account-42"), Value: []byte("credit")},
)
The example shows the key-based routing choice; configure the broker address, topic, and error handling for your application. Records with the same key are routed together by this balancer, while records with different keys can be spread across partitions. Ordering still applies only within each resulting partition.
Choose partitioning based on ordering and parallelism
| Design | Ordering scope | Consumer-group parallelism | When it fits |
|---|---|---|---|
| One partition | One sequence for the topic’s records | At most one group member actively reads that partition | Use when a single sequence matters more than partition-level parallelism. |
| Multiple partitions with stable key routing | Per key, provided the key consistently maps to one partition | Different partitions can be processed in parallel | Use for per-entity ordering while allowing concurrency among entities. |
| Unkeyed load balancing | Partition-local; related records can land in different partitions | Can distribute work, depending on the balancer | Use when distributing load matters more than preserving a shared sequence for related records. |
A consumer group assigns partitions among its members. Adding members can increase parallelism only while there are partitions available to assign; a group cannot have more active readers for a topic than that topic has partitions. Partition count is therefore both an ordering-design choice and a parallelism choice.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Preserve ordering when consumers process records concurrently
Fetching records from a partition in log order does not force application work to finish in that order. If a consumer dispatches records to concurrent workers, a later record can complete before an earlier one. That can violate business sequencing even though Kafka delivered the records in order.
Offsets are positions in a partition, not separate acknowledgment records for each message. kafka-go documents that committing the highest offset for a partition commits earlier offsets in that partition too. If a later task finishes first and its offset is committed while an earlier task is unfinished, a restart can skip that unfinished work.
Rank #4
In kafka-go consumer-group mode, ReadMessage commits offsets automatically. For explicit commit control, use FetchMessage and then CommitMessages. If processing concurrently, commit only when doing so is safe for all earlier records in that partition—for example, process sequentially per partition or track completed work and advance commits only through the contiguous completed sequence. See the kafka-go Reader documentation for the relevant APIs and behavior.
Quick Recap
Best Value
Practical decision checklist
- Identify what must remain in sequence: an entity’s events or every event in the topic.
- For per-entity ordering, use a stable entity key and explicitly configure a compatible key-based balancer.
- Choose enough partitions for the parallelism you need across keys, while recognizing that key-based ordering remains partition-local.
- Keep processing sequential where required, or coordinate completion and commits so no partition offset advances past unfinished work.
- For a single topic-wide sequence, use one partition and accept the limit on partition-level consumer-group parallelism.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
