Cost Based Optimization

The Denodo cost-based optimization process needs the statistics of at least all the base views involved in the query (the average size of a field, the maximum and minimum values, the number of distinct values, etc.) to estimate the cost of possible execution plans.

Statistics allow Denodo to decide whether to apply optimizations such as join reordering, choosing a join execution method (merge, hash, nested,…), applying ‘aggregation pushdown’, or performing data movement automatically.

It is therefore strongly recommended that for each base view created in Presto:

  1. Calculate statistics by calling the Denodo stored procedure COMPUTE_SOURCE_TABLE_STATS

  2. Gather statistics in Denodo

Add feedback