The automatic indexing feature automates index management tasks, such as creating, rebuilding, and dropping indexes in an Oracle Database based on changes in the application workload.
This feature improves database performance by managing indexes automatically in an Oracle Database.
SQL Diagnostics and Repair Enhancements
The SQL diagnostics and repair tools, such as SQL Test Case Builder and SQL Repair Advisor have been enhanced to provide better diagnosis and repair capabilities for managing problematic SQL statements.
These enhancements enable more effective diagnosis and repair of problematic SQL statements.
Bitmap Based Count Distinct SQL Function
Use new bit vector SQL operators to speed up COUNT DISTINCT operations within a SQL query. To compute COUNT(DISTINCT) for numeric expressions, you can create a bit vector representation of the expressions and aggregate them before the final bit count. The resulting bit vector can be materialized, such as in a materialized view.
You can construct bit vectors by further grouping on a larger set of GROUP BY keys than targeted queries, so that you can use one materialized view to rewrite multiple GROUP BY queries with COUNT(DISTINCT) expressions by using ROLLUP.
In most scenarios, bit vector SQL functions combined with materialized views can provide significant performance improvements for queries with COUNT(DISTINCT) operations, which are common in data warehousing environments. The new operators are naturally evaluated in parallel and take advantage of hardware-optimized bitmap operations. By creating materialized views with bit vectors at lower-level aggregation levels, you can reuse the same materialized view to rewrite queries at higher level of aggregation levels by using ROLLUP.
Big Data and Performance Enhancements for In-Memory External Tables
In-Memory external tables add support for ORACLE_HIVE and ORACLE_BIGDATA drivers, parallel query, Oracle Real Application Clusters, Oracle Active Data Guard, and on-demand population.
By using the new Big Data drivers, you avoid the cost and complexity of materializing data before populating it into the In-Memory Column Store (IM column store). You can use the SQL analytical capabilities of Oracle Database and Database In-Memory to analyze both internal and external data. Support for parallel query and full scan population means applications have fewer limitations when accessing data that resides outside the database.
Automatic SQL Plan Management
Automatic SQL plan management resolves plan regressions without user intervention. For example, if high-load statements are performing suboptimally, then SQL plan management evolve advisor can locate the statements automatically, and then test and accept the best plans.
SQL plan management searches for SQL statements in the Automatic Workload Repository (AWR). Prioritizing by highest load, it looks for alternative plans in all available sources, adding better-performing plans to the SQL plan baseline. Oracle Database also provides a plan comparison facility and improved hint reporting.
The impact of SQL statement performance regressions is significantly reduced using automation.
Oracle Database automatically gathers online statistics during conventional data manipulation language (DML) operations.
Statistics can go stale between execution of DBMS_STATS statistics gathering jobs. By gathering some statistics automatically during DML operations, the database augments the statistics gathered by DBMS_STATS. Fresh statistics enable the optimizer to produce more optimal plans.
You can configure a lightweight, high-frequency automatic task that periodically gathers optimizer statistics for stale objects.
Statistics can go stale between executions of DBMS_STATS jobs. By gathering statistics more frequently, the optimizer can produce more optimal plans.
Hybrid Partitioned Tables
The hybrid partitioned tables feature extends Oracle partitioning by enabling partitions to reside in both Oracle Database segments and in external files and sources. This feature significantly enhances the functionality of partitioning for Big Data SQL where large portions of a table can reside in external partitions.
Hybrid partitioned tables enable you to integrate internal partitions and external partitions into a single partition table. With this feature, you can also move non-active partitions to external files, such as Oracle Data Pump files, for a cheaper storage solution.