Understanding performance implications of INSERT INTO

[Siddharth Jain]

w.r.t. this:

First, it says 

• Add data in fairly large batches, such as 100,000 rows at a time.

then it says: Performance will not decrease if:

• Data is added in real time.

What is the meaning of real time here? To me, real time means given a stream of data (messages in Kafka e.g.), we  insert one row at a time as each message arrives. If that is the case, then the two statements are contradictory. Could anyone explain this to me? thanks.

[Siddharth Jain]

would the more appropriate statement be

Performance will not decrease if:

• Data is added in real time with the async_insert setting enabled?

[Antonio Rodriguez]

CH is built to works with good performance using 1 insert per second in a bulk operation with millions of rows in that insert.
No to insert 1 row per insert.