The maximum number of rows to fetch from Kafka at one time.

Data Type

int

Default Value

1000

Remarks

The provider batches reads to Kafka to reduce overhead. Instead of fetching a single row from the broker every time a query row is read, the provider will read multiple rows and save them to the resultset. This means that only the first row read from the resultset must wait for the broker. Later rows can be read out of this buffer directly.

This option controls the maximum number of rows the provider stores on the resultset. Setting this to a higher value will use more memory but requires waiting on the broker less often. Lower values will give lower throughput while using less memory.