Study Confluent CCAAK Topic Behavior: key concepts, common traps, and exam decision cues.
On this page
This lesson is where Kafka topic behavior stops being abstract. The exam is really asking whether the topic is meant to preserve event history, latest state, more parallelism, or tighter storage control.
Topic-intent chooser
Requirement
Strongest first fit
more consumer-group parallelism
more partitions
full replay window
delete retention policy with capacity planning
latest value per key
compaction
smaller storage footprint with event history trade-off
retention review
What the exam is really testing
If the scenario shows…
Strong reading
one entity must stay ordered
key and partition behavior matter
a changelog or latest-state stream
compaction is under test
storage growth
retention and segment policy may matter more than throughput tuning
more consumers than useful work
partition count may be limiting concurrency
Common traps
Trap
Better rule
expecting compaction to preserve full history
compaction optimizes latest value by key
increasing partitions without thinking about key distribution or ordering
more partitions change behavior, not just capacity
shrinking retention as the first response to any cluster stress
confirm the real constraint before cutting replay window
Decision order that usually wins
Ask whether the topic is meant to preserve history, preserve latest state, or unlock more concurrency.
If latest state by key matters, compaction is usually the first fit.
If replay depth matters, think retention and storage capacity before tuning other knobs.
If the stem asks for more consumer throughput, increase partitions only after considering ordering and key distribution effects.