The following features are new in this release:

• CDB fleet

  A CDB fleet is a collection of different CDBs that can be managed as one logical CDB.

  See "Administering a CDB Fleet".

• PDB snapshot carousel

  A PDB snapshot is a point-in-time copy of a PDB. When a PDB is enabled for snapshots, you can create multiple snapshots (point-in-time copies) of the PDB. The library of snapshots is called a PDB snapshot carousel. You can quickly clone a new PDB based on any snapshot in the carousel. In this way, you can perform point-in-time recovery to any snapshot in the carousel, or rapidly create a PDB by cloning any snapshot.

  See "User Interface for PDB Snapshot Carousel" and "Administering a PDB Snapshot Carousel".

• Logical partitioning

  A container map enables a session to issue SQL statements that are routed to the appropriate PDB, depending on the value of a predicate used in the SQL statement. The partitioning column in the map table does not need to match a column in the metadata-linked table. For example, if the table sales is enabled for the container map pdb_map_tbl, and if sales does not have the column used to partition pdb_map_tbl, then queries with the predicate CONTAINERS(sales) are still routed to the PDBs specified in the map table.

  See "Container Maps".

• Refreshable PDB switchover

  A refreshable clone PDB is a read-only clone that can periodically synchronize with its source PDB. You can reverse the roles, transforming the source PDB into the clone and the clone into the source. This technique can be useful for load balancing. Also, if the source PDB fails, then you can resume operations on the clone PDB, rendering a CDB-level Oracle Data Guard failover unnecessary.

  See "About Refreshable Clone PDBs" and "Switching Over a Refreshable Clone PDB".

• Lockdown profile enhancements

  You can create, alter, or drop lockdown profiles in application containers. Also, you can create lockdown profiles based on a static or a dynamic base profile.

  See "Overview of PDB Lockdown Profiles", "About Restricting PDB Users for Enhanced Security", and "Restricting Operations on PDBs Using PDB Lockdown Profiles".

• DBCA enhancements

  You can use DBCA to clone a local PDB or duplicate a CDB. Duplication is only supported in silent mode.

  See "About CDB Creation with DBCA" and "About Cloning a Local PDB".

• Usable backups of non-CDBs and relocated PDBs

  When you are cloning a non-CDB as a PDB or relocating a PDB, you can use the DBMS_PDB.EXPORTRMANBACKUP procedure to export RMAN backup metadata into the PDB dictionary. This metadata enables backups of the source non-CDB or PDB to be usable for restore and recovery of the target PDB.

  See "General Prerequisites for PDB Creation".

• RMAN duplication of a PDB to another CDB

  You can clone a PDB from a source CDB to an existing CDB that is open read/write.

  See "Techniques for Creating a PDB"

• Relocation of sessions during planned maintenance

  Application Continuity can drain database sessions during planned maintenance when the application submits a connection test, at request boundaries, and at good places to fail over. The relocation is transparent to applications. This feature is on by default for all maintenance operations invoked at the database service and PDB levels: stop service, relocate service, relocate PDB, and stop PDB.

  See "Managing Services for PDBs", "How PDB Relocation Works", and Oracle Real Application Clusters Administration and Deployment Guide.

• Copying a PDB in an Oracle Data Guard environment

  When performing a remote clone in a primary database, or plugging in a PDB in a primary database, you can set initialization parameters in a standby database that automates copying the data files for the newly created PDB.

  See "Cloning a Remote PDB or Non-CDB" and "Plugging In an Unplugged PDB".

• Parallel statement queuing at the PDB level

  You can configure parallel statement queuing for a PDB just as for a non-PDB using the PARALLEL_SERVERS_TARGET initialization parameter. At the PDB level, the default is based on the CPU_COUNT setting for the PDB. At the CDB level, the default value is the value of the PARALLEL_MAX_SERVERS initialization parameter.

  See "Utilization Limits for PDBs".

• Split mirror clone PDBs

  When a PDB resides in Oracle ASM, you can use a split mirroring technique to clone a PDB. The cloned PDB is independent of the original PDB. The principal use case is to rapidly provision test and development PDBs in an Oracle ASM environment.

  See "About Split Mirror Clone PDBs".

The following features are new in this release:

• Application containers

  An application container is an optional component of a multitenant container database (CDB) that consists of an application root and the application PDBs associated with it. An application container stores data for one or more applications.

  See "Overview of Applications in an Application Container".

• Application common objects

  Application common objects are created in an application root and are shared with the application PDBs that belong to the application root.

  See "Managing Application Common Objects".

• Support for thousands of pluggable databases (PDBs) in a single CDB

  A CDB can contain up to 4,096 PDBs.

• Use different character sets for PDBs

  When the character set of the CDB root is AL32UTF8, any container in the CDB can use a character set that is different from the CDB root and different from other containers in the CDB.

  See "Decide How to Configure the CDB" and Oracle Database Globalization Support Guide.

• Relocate a PDB from one CDB to another

  A PDB can be relocated in one operation with minimal down time.

  See "Relocating a PDB".

• Proxy PDB

  A proxy PDB references a PDB in a different CDB and provides fully functional access to the referenced PDB.

  See "Creating a PDB as a Proxy PDB"

• Hot PDB cloning

  The source PDB can be in open read/write mode during a PDB clone operation.

  See "Cloning a PDB or Non-CDB".

• Rename services during PDB creation

  The SERVICE_NAME_CONVERT clause of the CREATE PLUGGABLE DATABASE statement renames the user-defined services of the new PDB based on the service names of the source PDB.

  See "Service Name Conversion".

• Switch to a specific service for a container in a CDB

  You can specify a service name in an ALTER SESSION SET CONTAINER statement.

  See "Switching to a Container Using the ALTER SESSION Statement".

• Manage the memory usage of PDBs in a CDB

  You can configure guarantees and limits for SGA and PGA memory, using PDB initialization parameters.

  See "Memory-Related Initialization Parameters for PDBs."

• Limit the I/O generated by specific PDBs

  Two new initialization parameters, MAX_IOPS and MAX_MBPS, enable you to limit disk I/O generated by a PDB. MAX_IOPS limits the number of I/O operations, and MAX_MBPS limits the megabytes for I/O operations.

  See "I/O-Related Initialization Parameters for PDBs".

• PDB performance profiles

  You can specify Resource Manager directives for a set of PDBs using PDB performance profiles.

  See "Creating New CDB Resource Plan Directives for a PDB Performance Profile".

• Monitor PDBs managed by Oracle Database Resource Manager

  A set of dynamic performance views enables you to monitor the results of your Oracle Database Resource Manager settings for PDBs.

  See "Monitoring PDBs Managed by Oracle Database Resource Manager".

• Prioritize PDB upgrades

  You can prioritize the PDBs in a CDB when you upgrade the CDB. The PDBs with higher priority are upgraded before PDBs with lower priority.

  See Oracle Database Upgrade Guide.

• CDB undo mode

  A CDB can run in local undo mode or shared undo mode. Local undo mode means that every container in the CDB uses local undo. Shared undo mode means that there is one active undo tablespace for a single-instance CDB. For an Oracle RAC CDB, there is one active undo tablespace for each instance.

  See "Setting the Undo Mode in a CDB Using ALTER DATABASE".

• Parallelized PDB creation

  The PARALLEL clause of the CREATE PLUGGABLE DATABASE statement specifies whether to use parallel execution servers during PDB creation and, optionally, the degree of parallelism.

• Unplugging PDBs and plugging in PDBs with an archive file

  A PDB can be unplugged into compressed archive of the XML file that describes the PDB and the files used by the PDB (such as the data files and wallet file). The archive file has a .pdb extension, and it can be used to plug the PDB into a CDB or application container.

  "About Unplugging a PDB" and "About the XML File and Archive File".

• PDB refresh

  You can create a PDB as a refreshable clone and refresh the PDB with changes made to the source PDB.

  See "Refreshing a PDB".

• Improved support for default tablespace specification during PDB creation

  You can specify a default tablespace for a PDB that is created using techniques such as cloning and plugging in the PDB. Previously, a default tablespace could be specified only if the PDB was created from PDB$SEED.

  See "Default Tablespace".

• Extended USER_TABLESPACES clause during PDB creation

  The creation mode of user tablespaces can be different than the creation mode of the PDB. For example, during PDB creation, the user tablespaces can move a tablespace’s files even when file copy is specified for the PDB.

  See "User Tablespaces".

The following features are new in this release:

• Preserving the open mode of PDBs when the CDB restarts

  You can preserve the open mode of one or more PDBs when the CDB restarts by using the ALTER PLUGGABLE DATABASE SQL statement with a pdb_save_or_discard_state clause.

  See "Preserving or Discarding the Open Mode of PDBs When the CDB Restarts".

• The USER_TABLESPACES clause of the CREATE PLUGGABLE DATABASE statement

  You can use this clause to separate the data for multiple schemas into different PDBs. For example, assume that each schema in a non-CDB uses a separate tablespace. When you move a non-CDB to a PDB, and when the non-CDB has schemas that supported different applications, you can use this clause to separate the data belonging to each schema into a separate PDB,

  See "User Tablespaces".

• Excluding data when cloning a PDB

  The NO DATA clause of the CREATE PLUGGABLE DATABASE statement specifies that a PDB's data model definition is cloned but not the PDB's data.

• Default Oracle Managed Files file system directory or Oracle ASM disk group for a PDB's files

  The CREATE_FILE_DEST clause specifies the default location.

  See "PDB File Locations".

• Create a PDB by cloning a non-CDB

  You can create a PDB by cloning a non-CDB with a CREATE PLUGGABLE DATABASE statement that includes the FROM clause.

  See "Cloning a PDB or Non-CDB".

• The logging_clause of the CREATE PLUGGABLE DATABASE and ALTER PLUGGABLE DATABASE statement

  This clause specifies the logging attribute of the PDB. The logging attribute controls whether certain DML operations are logged in the redo log file (LOGGING) or not (NOLOGGING).

  See "Modifying a PDB at the Database Level" for information about this clause and the ALTER PLUGGABLE DATABASE statement.

• The pdb_force_logging_clause of the ALTER PLUGGABLE DATABASE statement

  This clause places a PDB into force logging or force nologging mode or takes a PDB out of force logging or force nologging mode.

  See "Modifying a PDB at the Database Level".

• The STANDBYS clause of the CREATE PLUGGABLE DATABASE statement

  This clause specifies whether the new PDB is included in standby CDBs.

  See Oracle Data Guard Concepts and Administration for more information about configuring PDBs on standby CDBs.

• Querying user-created tables and views across all PDBs

  The CONTAINERS clause enables you to query user-created tables and views across all PDBs in a CDB.

  See Querying User-Created Tables and Views Across All Containers.

The following features are new in this release:

• Oracle Multitenant option

  Oracle Multitenant option enables an Oracle database to function as a multitenant container database (CDB) that includes one or many customer-created pluggable databases (PDBs). A PDB is a portable collection of schemas, schema objects, and nonschema objects that appears to an Oracle Net client as a non-CDB. All Oracle databases before Oracle Database 12c were non-CDBs. You can unplug a PDB from a CDB and plug it into a different CDB.

  See "Administering a Multitenant Environment".

• Resource Manager support for a multitenant environment

  Resource Manager can manage resources on the CDB level and on the PDB level. You can create a CDB resource plan that allocates resources to the entire CDB and to individual PDBs. You can allocate more resources to some PDBs and less to others, or you can specify that all PDBs share resources equally.

  See "Using Oracle Resource Manager for PDBs".