Hacker News Reader

I think your missing my point. The page is entitled "a Benchmark For Analytical DBMS" not "A Benchmark for Single Table Query Execution". Most analytical workloads are more complex then single table queries.

I didn't say it wasn't useful to test single table columnstore performance on workload that runs best on single host databases, just that this isn't the be-all end-all of Analytical Database performance testing.

You are absolutely right. That's why this benchmark is named "a Benchmark For Analytical DBMS", not "the definitive benchmark for analytical DBMS".

There's a lot more involved in an execution engine running complex queries that are not single table group by than just QO (though this is important). It includes things like join implementations and associated optimizations, shuffle performance (which is important even for single table queries as you scale), etc.

> It does not show performance of query engine, but more likely how good is your optimiser was tailored for this queries.

you can join just two large tables without leaving much space for query optimizer.