Wednesday, 14. October 2015., 12:30
Hall 2 Castle
Based on object statistics stored in the data dictionary, the query optimizer can’t always accurately estimate the cost of an operation or predicate. When the query optimizer recognizes such a case, in some situations it can gather additional statistics through dynamic sampling. The original implementation of dynamic sampling was introduced in Oracle Database 9.2 and, up to 220.127.116.11, only marginally improved. However, Oracle Database 18.104.22.168 and 12.1 introduces a new implementation that completely changes the way dynamic sampling works. The aim of this presentation is to explain and demonstrate how the new implementation works.