When did asynchronous data inserts become the default in ClickHouse?

ClickHouse consciously adopted asynchronous inserts as the default mode starting with the version 26.3 release in April 2026.

What happens if wait_for_async_insert is set to 0?

When wait_for_async_insert is set to 0, the client receives an acknowledgment immediately after data is written to the in-memory buffer, before it is flushed to disk. This dramatically reduces latency but carries a small risk of data loss if the server crashes.

How do synchronous inserts impact ClickHouse performance?

Synchronous inserts provide maximum durability, but because every insert creates a new part on disk, sending many small inserts can overwhelm the server and cause 'Too many parts' errors, degrading query performance.

What triggers a data flush when using asynchronous inserts?

ClickHouse triggers a flush to disk when any one of three conditions is met: the timeout limit (async_insert_busy_timeout_ms), the maximum data size (async_insert_max_data_size), or the maximum query number (async_insert_max_query_number).