Pluggable Database Primer - Protecting Oracle Database 12c - page 60

CDB

PDB

PDB

PDB

Shared processes

Figure 7-1. Basic CDB container with pluggable databases in multi-tenancy

The architecture for achieving this is a single central CDB with multiple PDB databases on top. The PDBs are the
individual consolidated databases and are logically isolated—but not physically in practice.
The CDB contains:
-CDB$ROOT , which is the master dictionary but with no data as such
-PDB$SEED , used to create new PDBs so it does have a dictionary

-0 to 252 PDBs
-CDB_USERS view , which contains new column CON_ID with root CON_ID=1 for
CDB$ROOT

Note:

-Each PDB has a separate system tablespace and can have a different time zone.

-All PDBs use the same UNDO tablespace.
-There is one common alert.log and set of trace files for the CDB.

-There is a common spfile, control file, and SGA for all PDBs of a single CDB.
-PDB parameters are viewable from v$system_parameter with the con_id of the PDB.

-You cannot enable flashback at pluggable DB level.

-PDBs can have their own tablespace (in fact, this is recommended).
- V$PARAMETER.ISPDB_MODIFIABLE shows which parameters are modifiable within the
instance without requiring a restart.

-PDBs and CDB have to be the same version of the DB and have the same character set.

The key point when starting to work with containers is that starting the CDB database does not automatically
open the pluggable databases contained within:

$ sqlplus / as sysdba

SQL> startup;

You then need to open all pluggable databases with this command:

SQL> alter pluggable database all open;