Similarly as we did in db-benchmark 1.0.0 for Impala and Presto SQL databases, we should do for ClickHouse.

In short, all the other solutions are capable to keep answers of their queries in a variable (cache=TRUE). On the other hand results of Clickhouse queries are immediately lost (cache=FALSE). Field cache is in time.csv for that purpose, so we can run both scenarios if needed.
This issue requests measuring queries with cache=TRUE for ClickHouse, which will need to wrap current queries into CREATE TABLE AS SELECT, so the results are stored in a table.