Redo Log Files
Whenever data is added, removed, or changed in a table, index, or other Oracle object, an entry
is written to the current redo log file. Every Oracle database must have at least two redo log files,
because Oracle reuses redo log files in a circular fashion. When one redo log file is filled with
redo log entries, the current log file is marked as ACTIVE, if it is still needed for instance recovery,
or INACTIVE, if it is not needed for instance recovery; the next log file in the sequence is reused
from the beginning of the file and is marked as CURRENT.
Ideally, the information in a redo log file is never used. However, when a power failure
occurs, or some other server failure causes the Oracle instance to fail, the new or updated data
blocks in the database buffer cache may not yet have been written to the datafiles. When the
Oracle instance is restarted, the entries in the redo log file are applied to the database datafiles
in a roll forward operation, to restore the state of the database up to the point where the failure
occurred.
To be able to recover from the loss of one redo log file within a redo log group, multiple
copies of a redo log file can exist on different physical disks. Later in this chapter, you will
see how redo log files, archived log files, and control files can be multiplexed to ensure the
availability and data integrity of the Oracle database.
Control Files
Every Oracle database has at least one control file that maintains the metadata of the database
(in other words, data about the physical structure of the database itself).
Pages:
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99