| ETERNUS SF Recovery Manager for Oracle EE 13.0.1 User's Guide - Solaris(TM) Operating System - |
|
Contents
|
ARCHIVELOG is the redo log mode archived before the online redo log files are reused. In the ARCHIVELOG mode, the database can be recovered fully from instance or disk failures. While the database is open and is in a usable state, it can be backed up. However, the archived redo log must be maintained through additional management operations.
Archived log mode is one of the database modes.
In ARCHIVELOG mode, the log transfer service archives on the disk the online redo logs that have filled up. The mode can be specified at the time of database creation, or using the "ALTER DATABASE ARCHIVELOG" SQL statement.
To enable the auto archive mode, use the "ALTER SYSTEM ARCHIVE LOG START" SQL statement dynamically or specify "true" in the initial parameter" LOG_ ARCHIVE_START".
A copy of data such as database, tablespace, table, data file, control file, and archived redo log
Backup data can be created as follows:
The Oracle Recovery Manager (RMAN) command for creating a backup set
A backup set is a backup data of one or more data files, control files or archived logs generated by the "BACKUP" command of the Oracle Recovery Manager (RMAN).
A backup set is a logical group consisting of one or more binary files called backup pieces. A backup set is created in a dedicated format and can be restored only by Oracle Recovery Manager (RMAN).
A type of incomplete media recovery operation.
To use the cancel-based recovery function, specify the "UNITL CANCEL" clause with the RECOVER" command. The recovery is continued until the "CANCEL" command is issued.
A Change-based recovery is a type of the incomplete media recovery that recovers data up to a specified SCN.
It is also possible to perform a cancel-based recovery operation that recovers data up until the issue of the CANCEL command, or a time-based recovery operation that recovers data up to a specified time.
A cluster system is a highly available system connecting multiple nodes via a dedicated network.
High availability is achieved by inheriting transactions through switching nodes automatically when a node is down.
The complete recovery operation recovers one or more data files by applying all the online redo and archived redo logs generated after the restored backup operation.
The complete media recovery operation is normally executed when one or more data files or the control file is damaged through an error with the medium.
To recover the damaged file completely, all the redo logs generated after the restored backup operation are used.
When the Oracle Recovery Manager (RMAN) is in use, incremental backup can also be applied at the time of the complete recovery operation.
A control file is a binary file associated with a database. The physical structure and time stamp are maintained for all files in the same database. The database server continually updates the control files while the database is being used. When the database is mounted or open, the database server must always be able to write to the control files.
A data file is a physical operating system file located on the disk that has been created by the database server, and includes tables, indexes, and other structured data. A data file can belong to only one database.
An advanced copy function, EC (Equivalent Copy), copies data in the disk partition to a different disk slice, and maintains the original and the destination copy in the equivalent state.
One of the interfaces for serial data transfers. High-speed transfer and long distance transmission to and from servers are the characteristics of the fiber channel.
The PRIMECLUSTER GDS is volume management software that mirrors the disk device locating systems and data, and protects clients' assets from unexpected events such as a disk failure. It supports the concatenation feature that makes possible the creation of large capacity volumes, and the striping feature that distributes access load on the disk.
The PRIMECLUSTER GFS is a file system with excellent extendibility, availability, and performance. It provides functions such as online size extension, high-speed recovery, and contiguous block allocation. In the cluster environment, it makes the file system possible on the shared storage unit.
In an incomplete recovery operation, the database is recovered without applying some of the changes generated after creating the restoring backup.
Incomplete recovery is normally executed in the following instances:
The database is recovered to the state of the latest archived log generated prior to the hardware failure.
The data must be recovered to a point prior to the occurrence of an inappropriate action generated in the database.
For example, if a user accidentally deleted a transaction for salary payment (before it is sent to the salary dispensing institution), the database administrator must restore the entire database to the point immediately before the transaction deletion by executing the incomplete media recovery.
The only solution is to recover the data to the point when the log was lost.
In either case, after recovering the medium, open the database with "RESETLOGS" option.
The initialization parameter files, such as "spfile.ora", "initdbname.ora", and "initsid.ora", are information files required to initialize databases.
An instance is a combination of database processes and the system global area (SGA). The memory and the process of an instance manage the data in the relevant database and function for the database users.
An instance has the unique Oracle system identifier, "INSTANCE_NAME", "INSTANCE_NUMBER", roll back segment, and thread number.
An independent process executed by a server. The listener listens to a request for reception and connection from a client, and manages the amount of traffic to the server. The listener is a component of Oracle Net, an Oracle remote data access product.
The log sequence number is a number used to uniquely identify a set of redo records in redo log files. When an online redo log file becomes full and the file is switched to another online redo log file, Oracle automatically allocates a log sequence number to the new file. For example, if a database has been created with two online redo log files, Oracle allocates 1 to the first file as the log sequence number. When this file becomes full, Oracle switches to the next file and allocates 2 as the log sequence number. When Oracle switches to the first file again, it allocates 3 as the log sequence number.
Log sequence number recovery is a type of incomplete recovery executed by Oracle Recovery Manager (RMAN). It recovers data up to the specified log sequence number.
Log switch occurs when LGWR stops writing to the active redo log and switches to the next usable redo log. LGWR switches the log when the active log has become full with the redo records, or when switching is forced manually. When the database operation is executed in ARCHIVELOG mode, the log transfer service archives the inactive log redo data in the archived redo log. When the log is switched and LGWR writes over the previous redo data (because the archived redo log contains the previous data) the data will not be lost. In NOARCHIVELOG mode, the log transfer service writes over the previous data without archiving it when the log is switched. Therefore all the old redo data will be lost.
An LU is a unit of logical disk specified from the server.
An LUN is a logical unit number defined under the channel adapter in the disk array.
A media failure is a physical cause for Oracle to fail to read or write files required for database operation. A common example is crashing of the disk head resulting in a loss of all data on the disk drive. A disk error can affect many kinds of files such as data files, redo log files, and control files. After a media failure, the database instance no longer operates normally. Data in the buffer cache in the SGA volume cannot be written to a data file.
Media recovery applies redo or incremental backup data, and returns the restored backup data files or individual data blocks to the state as of a specified time. A Media recovery of data files always starts from the lowest SCN recorded in the data file header.
The following can be recovered by executing a media recovery:
Normally, the media recovery is executed when part or all of the database files (data files, control files, or online redo logs) are damaged due to a media error. In the ARCHIVELOG mode, either complete or incomplete recovery can be selected. In NOARCHIVELOG mode, data can normally be restored only from the latest backup data without applying the redo data.
The online redo log records all the changes made to the data files and control files. It is a set consists of two or more files.
Every time the database is changed, Oracle generates the redo records on the redo buffers. The contents of the redo buffers are flushed to the online redo log by the LGWR process.
The current online redo log refers to the online redo log that is currently written to by LGWR.
When LGWR reaches the end of a file, it switches the log and starts writing to a new log file.
When the database is in the ARCHIVELOG mode, redo data is copied to the archived redo log by one or more archived process.
This function enables high-speed backup and restore operations while running a transaction. It also enables the generation management of backup data. The OPC (One Point Copy) advanced copy function copies at high speed a group of contiguous blocks on the LUN in storage, to an area on a different LUN in the same disk array.
A RAW device processes I/O operations directly without going via the caching layer.
Normally, recovery refers to an operation to repair a database failure.
The recovery catalog refers to a set of tables and views Oracle Recovery Manager (RMAN) uses to locate information on a database. RMAN manages backup, restore, and recovery operations of Oracle databases using this information. When no recovery catalog is used, RMAN uses the control files of the target database. The recovery catalog may not be located in the target database.
A recovery catalog database locates the schema of the recovery catalogs.
Normally, restore means to return backed up data using the export utility etc. to the database.
Resynchronization processing updates recovery catalogs using the current control file information of the target database. Complete catalog resynchronization can be initiated by issuing the "RESYNC CATALOG" command. In the partial resynchronization, information related to copying of the archived redo log, backup sets and data files are transferred to the recovery catalog. Oracle Recovery Manager (RMAN) executes resynchronization automatically as required.
The Oracle Recovery Manager is a utility that executes the backup, restore, and recovery operations of databases. It can be used in conjunction with, or without, the central information repository called the recovery catalog.
A SAN (Storage Area Network) is a network for connecting servers and storages. Broadly defined, it represents a comprehensive concept that covers storage concentration, data sharing among different servers, and the implementation of storage integrated management. It is described as technology for the future.
Currently, a configuration in which servers and storage are connected using fiber channels like a network is starting to be implemented. Commonly, this configuration is referred to as the "SAN". Presently, sharing of files between different operation system types is not possible.
The System Global Area (SGA) refers to the area in main memory that locates data for high-speed access operations.
Oracle uses the shared pool for shared SQL and the memory allocation of SGA memory for PL/SQL procedures.
Storage is a generic term for storage devices such as floppy disk drives, hard disks, MO(magneto-optical), and other devices that locate data. It includes external storage devices, storage systems, and tape devices.
Every time a transaction is committed, the SCN (system change number) is issued to the transaction. SCN is a unique identifier acting as the internal time stamp of the database. Normally, the value of SCN is simply incremented by one, but it is guaranteed not to overflow for 500 years at 16000tps. The database maintains integrity using the SCN values.
In the Oracle RDBMS, a logical storage area is created to locate tables and indexes. The storage area is associated with the files on the disk (physical structure). This improves the efficiency of large-scale data management. This logical storage area in Oracle RDBMS is referred to as the "tablespace". A logical structure of the storage area is data block < extent < segment < tablespace.
The target database refers to the database backed up or restored by Oracle Recovery Manager (RMAN).
The time-based recovery operation performs incomplete recovery of the database files up to a time other than the current time. It is also called the point-in-time recovery operation.
Two types of operation are available
Subsequently, the tablespace is re-integrated into the original database.
The Unix File System is a group of files that can be mounted to the block special files on Unix. Files comprising the file system appear on the i list of the file system only once, and can be accessed from the root directory of the file system via a specific path.
v$ table / dynamic performance views are the views created on the dynamic performance table (virtual table recording of the current database activities) by the database administrator.
Since it cannot be changed or deleted by the database administrator, it is referred to as the fixed view.
|
Contents
|