>_

Kafka Message Flow

Producers append to partitions; a consumer group pulls each partition with one consumer.

Produced

0

Consumed

0

In flight

0

Partitions

4

Producer Partition Consumer In flight
appendpullPProducer0part1part2part3part0consumer1consumer

Partition → consumer ownership

C00, 2
C11, 3

Each message goes to one partition (round-robin), then to that partition’s single consumer. A partition has exactly one consumer; a consumer may own several. The amber dots are in flight — watch the +n counts climb as flow moves left → right.

How this simulator works

Producers append to partitions; a consumer group pulls each partition with at most one consumer. Watch messages flow producer → partition → consumer.

Partitions and parallelism

  • A partition is the unit of parallelism — one partition is consumed by exactly one consumer in a group.
  • Adding consumers past the partition count leaves consumers idle.
  • Adding partitions lets you scale consumption (and rebalances the group).

Ordering

Ordering is guaranteed within a partition, not across partitions. Keyed production keeps a key’s events on one partition and therefore ordered.