Which statement about co-partitioning requirements for KStream-to-GlobalKTable joins is true?

Which statement about co-partitioning requirements for KStream-to-GlobalKTable joins is true?

• A. Yes
• B. No
• C. Only if both topics are windowed
• D. No, not required for GlobalKTable joins

Answer: (D)

Co-partitioning is about ensuring that the same key from two streams lands on the same partition so a join can be performed without cross-node lookups. For a KStream-to-GlobalKTable join, you don’t need this alignment because the GlobalKTable is replicated to every instance, giving each worker a local copy of the entire table. That means a KStream record can be joined with its corresponding GlobalKTable entry on the same node regardless of partition, so co-partitioning isn’t required. Windowing or lack thereof doesn’t change this fact; the join can happen correctly without partition alignment.

Co-partitioning is about ensuring that the same key from two streams lands on the same partition so a join can be performed without cross-node lookups. For a KStream-to-GlobalKTable join, you don’t need this alignment because the GlobalKTable is replicated to every instance, giving each worker a local copy of the entire table. That means a KStream record can be joined with its corresponding GlobalKTable entry on the same node regardless of partition, so co-partitioning isn’t required. Windowing or lack thereof doesn’t change this fact; the join can happen correctly without partition alignment.