DBAs and Query Optimisers – dive into the invisible part of the Columnstore Indexes.
Every time you see a Columnstore Index getting involved in the execution plan, do you realise that there are whole execution plans behind those Index Scans ? Did you ever ask yourself, what are those strange and weird HT_* waits stand for ? Why do we wait for seconds/minutes for something like HTBUILD while it seems that nothing happens ? Why do we have a ROWGROUP_VERSION wait on one server, while the other allows queries to run faster ?
This session focuses on answering those question – to help you understand the reasons and the conditions behind every single available wait for the Columnstore Indexes and the Batch Execution Mode.
Why I Want to Present This Session:
There is no real official documentation on the topic of the Columnstore Indexes waits and I believe it is a geeky fun to dive into the matter and understand the implementation as closely as possible to the real code. A lot of times people will see the Waits but will have no idea what they mean or how they can potentially tune them – this session should provide the insights.
Latest posts by Niko Neugebauer (see all)
- Columnstore Clustered vs Nonclustered, when to choose what - June 11, 2018
- Columnstore Indexes: What are we waiting for? - May 10, 2018
- Worst Practices & Less Known Limitations for Columnstore Indexes - April 21, 2017