There
are three major sets of files on disk that compose a database. All the files
are binary. These are
Database files
Control files
Redo logs
The most important of these
are the database files where the actual data resides. The control files and the
redo logs support the functioning of the architecture itself.
All
three sets of files must be present, open, and available to Oracle for any data
on the database to be useable. Without these files, you cannot access the
database, and the database administrator might have to recover some or all of
the database using a backup, if there is one.
Database Files
The database
files hold the actual data and are typically the largest in size. Depending on
their sizes, the tables (and other objects) for all the user accounts can go in
one database file—but that's not an ideal situation because it does not make
the database structure very flexible for controlling access to storage for
different users, putting the database on different disk drives, or backing up
and restoring just part of the database.
You must have at least one
database file but usually, more than one files are used. In terms of accessing
and using the data in the tables and other objects, the number (or location) of
the files is immaterial.
The database files are fixed in size and never grow
bigger than the size at which they were created.
Control Files
The control files and redo logs support the rest of the
architecture. Any database must have at least one control file, although you
typically have more than one to guard against loss. The control file records
the name of the database, the date and time it was created, the location of the
database and redo logs, and the synchronization information to ensure that all
three sets of files are always in step. Every time you add a new database or
redo log file to the database, the information is recorded in the control
files.
Redo
Logs
Any database
must have at least two redo logs. These are the journals for the database; the
redo logs record all changes to the user objects or system objects. If any type
of failure occurs, the changes recorded in the redo logs can be used to bring
the database to a consistent state without losing any committed transactions.
In the case of non-data loss failure, Oracle can apply the information in the
redo logs automatically without intervention from the DBA. The redo log
files are fixed in size and never grow dynamically from the size at which they
were created.
No comments:
Post a Comment