E4 2016 in Barcelona – Day 1

First day starting with “Extending Database In-Memory on Exadata Flash” by Roger MacNicol from Oracle. Very Good presentation!
Followed by SQL-on-Hadoop for Analytics and Data Integration – What are My Options, What is the Future? by Mark Rittman.
Oracle Exadata and Database Memory – Frits Hoogland of Accenture Enkitec Group
Next, SuperCluster M7 – Configuration Guidance and Best Practices by Oracle. OVM server for SPARC has the virtualization technology built into the M7 chip. For more than four domains, we setup root domains and IO domains.

Oracle Extended Statistics

Introduced 11gR1 and based on virtual columns. Allows better selectivity.

Condition has the be equality or IN-lists.

We should use them when we have issues by an order of magnitude on the cardinality.

Types:

  • Column Group: Have to be from same table. Max 32
  • Expression: CBO does not know how an applied function impacts the cardinality of a column so it uses default of 1%!!!

They can be seen in DBA_TAB_COLS by checking the hidden columns. Or in data dictionary by checking *_STAT_EXTENSIONS. They have system generated names. The 12c prefix is different of the 11g prefix.

Question: how to fond the correlations?

  • 11gR1 manually
  • 11gR2 manually / assisted
  • 12gR1 manually / assisted / automatic