This chapter contains:
The following are the changes in Oracle Database Data Warehousing Guide for Oracle Database 12c Release 1 (188.8.131.52).
Oracle In-Memory Column Store
The Oracle In-Memory Column Store (IM column store) in an optional area in the SGA that stores tables, table partitions, and individual columns in a compressed columnar format. The IM column store is a supplement to rather than a replacement for the database buffer cache.
The IM column store primarily improves the performance of table scans and the application of
WHERE clause predicates. Faster table scans make it more likely that the optimizer will choose bloom filters and
VECTOR GROUP BY transformations.
Attribute clustering of tables enables you to store data in close proximity on disk in a ordered way that is based on the values of certain columns in the table. I/O and CPU costs of table scans and table data lookup through indexes are reduced because pruning though table zone maps becomes more effective.
Zone maps enable natural pruning of data based on physical location of the data on disk. Accessing only the relevant data blocks during full table scans and accessing only the relevant data rows during index scans reduces I/O and CPU costs of data access.
VECTOR GROUP BY operation improves the performance of queries that join one or more relatively small tables to a larger table and aggregate data. In the context of data warehousing,
VECTOR GROUP BY aggregation will often be chosen for star queries that select data from the IM column store.
VECTOR GROUP BY aggregation minimizes the processing involved in joining multiple dimension tables to one fact table. It uses the infrastructure related to parallel query and blends it with CPU-efficient algorithms that maximize performance.
Automatic Big Table Caching
Automatic big table caching improves in-memory query performance for large tables that do not fit completely in the buffer cache. Such tables can be stored in the big table cache, an optional, configurable portion of the database buffer cache.
The following are changes in Oracle Database Data Warehousing Guide for Oracle Database 12c Release 1 (184.108.40.206).
SQL has been extended to support pattern matching, which makes it easy to detect various patterns over sequences. Pattern matching is useful in many commercial applications, such as stock monitoring, network intrusion detection, and e-commerce purchase tracking.
" SQL for Pattern Matching " for more information
Native SQL Support for Top-N Queries
row_limiting_clause enables you to limit the rows returned by a query. You can specify an offset, and number of rows or percentage of rows to return. This enables you to implement top-N reporting.
"Limiting SQL Rows" for more information
Online Statistics Gathering for Bulk Load Operations
Starting in Oracle Database 12c, the database automatically gathers table statistics as part of bulk load operations.
A new type of refresh called synchronous refresh enables you to keep a set of tables and materialized views defined on them to always be in sync. It is well suited for data warehouses where the loading of incremental data is tightly controlled and occurs at periodic intervals.
"Synchronous Refresh" for more information
A new type of refresh is available to improve materialized view refresh performance and availability. This refresh, called out-of-place refresh because it uses outside tables during refresh, is particularly effective when handling situations with large amounts of data changes, where conventional DML statements do not scale well.
"Refreshing Materialized Views" for more information
Some features previously described in this document are desupported in Oracle Database 12c Release 1. See Oracle Database Upgrade Guide for a list of desupported features.