Currently columnstore does not have explicit query optimization phase - the optimization process is spread across several modules, some of them rewrite the execution plan (using peephole optimization), some of them rule out some of the operations of execution plan (extent elimination is one example). And, while we have "analyze_table" query, currently we do not have necessary means to use its results in a straightforward way in a planning , let alone use them with ease.

The goals for this particular task are twofold:

1. create a framework for query optimization stage that can be easily integrable with various parts of columnstore's internal information and
2. demonstrate its effectiveness on different queries of various complexity - effectiveness in the addition of transformations and effectiveness of the resulting execution plan.