Database Related Files
Oracle database is a collection of physical
files. These are the operating system files used
by the database and database instance. Oracle
RAC needs shared storage to store the files. RAC
follows the shared disk model where all the
cluster nodes share the same disk or storage
This is called a “shared-everything”
The files included in the RAC architecture are
shown in Figure 2.4. Most of them must be
available simultaneously and be updateable by
all the nodes/instances in the cluster. Some
files can remain on the local file system.
Database Related Files
These .dbf files are the main files of the
database. The data files contain the actual
data. The logical database structures like
tables and indexes are physically stored in the
data files. In RAC, these files are located on
shared storage and are accessible by all the
nodes in the cluster. One or more data files
form a logical unit of database storage called
Data files can be associated with multiple
instances but only one database. By locating the
data files either on a clustered file system, a
network file system, or a raw partition, they
are made accessible by all the nodes.
The control files contain entries that specify
the physical structure of the database. This
small binary file is continuously updated when a
database instance is online.
The control files contain the key
information such as the database name, name and
location of the data files, and redo log files
for the database.
When an instance is launched, the control files
identify the data and redo log files. Control
files should be multiplexed and located on the
Redo Log Files
Oracle defines the redo log as the most crucial
structure for recovery operations.
A redo log is made up of redo entries
that are also called redo records. The primary
function of the redo log is to record all
changes made to data.
In a high update database, moving the
redo logs to separate disks is advised.
Every database has a set of redo log
files. The information in the redo log files is
used to recover the database from a system or
media failure. There are generally two or more
redo log files. They are used by the database in
a circular fashion. Once a redo log file is
filled up, then the next redo log file is picked
up for writing. Meanwhile, the filled redo log
file is saved as an archived log file.
Redo log files are stored as a group called Redo
Log File groups. Each group can have one or more
redo log files. Multiplexing the redo files
within a group provides a higher level of
resiliency and job security.
Redo log files are instance specific. In the RAC
database architecture, each instance has its own
set of redo log file groups. Even though they
are specific to an individual instance, the redo
log files need to be located on shared storage
for recovery purposes.
Another important use of redo logs is hot
mining of redo log files by Oracle streams where
redo log files are scanned in order to propagate
the changes to other Oracle database systems.