Performing Oracle User-Managed Backups - OCP: Oracle Database 12c Administrator Certified Professional - page 19

shutdown processes. We will then discuss the mechanics of actually starting and stopping
the database in its different phases. We will quickly discuss restricted-mode database oper-
ations, and finally we will discuss the different stages the database must be in for specific
types of backup and recovery operations to occur.
Exploring the Stages of Database Startup and Shutdown
When an Oracle database is started, it goes through four different and distinct stages:
Shutdown When the database and instance are shut down, they are at rest. There are no
processes present, no memory is allocated, and nothing is going on. It is important to note
that even though the database/instance may be shut down and closed, other Oracle processes
(like the listener or OEM agents) may be still running.
Nomount When the database is in nomount mode, the instance has been started. Thus,
processes have been started and memory allocated.
Mount When the database is in mount mode (or mounted), the instance is started and the
database has opened the control file. The control file is read, but its contents are not validated.
Note that the database is not open at this time.
Open When the database is opened, the control file contents have been validated
against the physical database. The data files are all confirmed to be present, and they
are opened. Oracle will then analyze the data files to determine if the database is in a
consistent state. If the database is not in a consistent state, some form of recovery will
be required.
Typically, the form of recovery required, crash or instance recovery , does not require any
DBA involvement. If instance recovery is not possible, then media recovery is required. Media
recovery requires the application of backups and recovery operations to bring the database
current to the point of failure (if this is possible). The principal determining factors for media
recovery are the presence of the needed data files and the availability in the online redo logs
of the redo needed to bring those files current. If either of these conditions does not exist,
then media recovery is required.
Database shutdowns occur in much the same way as startups, except in reverse. There
are two different kinds of shutdowns, however: consistent and inconsistent.
Consistent Shutdowns If your database shutdown is a consistent shutdown , then the data-
base data files and the database control file will be synchronized upon shutdown. The dirty
buffers in the database buffer cache will be flushed out to the database data files, making
them consistent. A consistent shutdown is a nice, tidy shutdown.
Inconsistent Shutdowns An inconsistent shutdown is another term for a mess. When your
database is shut down in an inconsistent manner, it is in an indeterminate state and will
require some form of recovery (typically instance recovery, which requires no DBA interven-
tion) when it is restarted. Inconsistent shutdowns, however bad they might sound, often are
the only way to shut down a database in a timely manner.