一.官网说明

Using Oracle Managed Files

http://docs.oracle.com/cd/E11882_01/server.112/e25494/omf.htm#ADMIN003

1.1 What Are Oracle ManagedFiles?

官网链接：

http://docs.oracle.com/cd/E11882_01/server.112/e25494/omf001.htm#i1006122

Using Oracle Managed Files simplifies the administrationof anOracle Database. Oracle Managed Files eliminate the need foryou, the DBA, todirectly manage the operating system files thatcomprise an Oracle Database.With Oracle Managed Files, you specifyfile system directories in which thedatabase automatically creates,names, and manages files at the database objectlevel. For example,you need only specify that you want to create a tablespace;you donot need to specify the name and path of the tablespace's datafilewiththe DATAFILE clause. Thisfeature works well with a logical volumemanager (LVM).

--使用OMF可以简化Oracle数据库的管理，通过OMF，可以指定filesystem 目录，之后就可以在这个目录下创建和管理数据库。比如创建表空间时，不需要指定datafile的名称，默认会将数据文件创建到这个目录下面。

The databaseinternally uses standard file system interfaces tocreate and delete files asneeded for the following databasestructures:

--Oracle默认使用标准的文件系统接口来创建和删除如下的数据库结构：

（1）Tablespaces

（2）Redo log files

（3）Control files

（4）Archived logs

（5）Block change tracking files

（6）Flashback logs

（7）RMAN backups

Throughinitialization parameters, you specify the file systemdirectory to be used fora particular type of file. The databasethen ensures that a unique file, anOracle managed file, is createdand deleted when no longer needed.

--在初始化参数里指定特殊类型文件的文件系统目录，这样就可以保证文件的唯一性。

This featuredoes not affect the creation or naming ofadministrative files such as tracefiles, audit files, alert logs,and core files.

1.2 WhoCan Use Oracle ManagedFiles?

Oracle Managed Files are most useful forthe following types ofdatabases:

--OMF 在如下情况使用最多：

（1）Databases that are supported by thefollowing:

1）A logical volume manager thatsupports striping/RAID anddynamically extensible logical volumes

2）A file system that provides large,extensible files

（2）Low end or test databases

The OracleManaged Files feature is not intended to easeadministration of systems thatuse raw disks. This feature providesbetter integration with operating systemfunctionality for diskspace allocation. Since there is no operating systemsupport forallocation of raw disks (it is done manually), this featurecannothelp. However, because Oracle Managed Files require that youuse the operatingsystem file system (unlike raw disks), you losecontrol over how files are laidout on the disks and thus, you losesome I/O tuning ability.

--OMF 没有简化raw disks的管理，OMF提供了一个整体的磁盘空间分配系统。因为OMF需要使用OS文件系统，所以可以丢失了一些对磁盘的控制，比如具体的文件存放那个磁盘，I/O性能调整的。

1.3 WhatIs a Logical VolumeManager? –什么是LVM

A logical volume manager (LVM) is a softwarepackageavailable with most operating systems. Sometimes it iscalled a logical diskmanager (LDM). It allows pieces of multiplephysical disks to be combined intoa single contiguous address spacethat appears as one disk to higher layers ofsoftware. An LVM canmake the logical volume have better capacity,performance,reliability, and availability characteristics than anyof the underlyingphysical disks. It uses techniques such asmirroring, striping, concatenation,and RAID 5 to implement thesecharacteristics.

--LVM 在大多数系统下都可以用，LVM 也称LDM。LVM可以让逻辑卷有更好的容量，性能和可靠性。LVM可以使用mirror，strip，concatenation和RAID技术。

Some LVMs allowthe characteristics of a logical volume to bechanged after it is created, evenwhile it is in use. The volume maybe resized or mirrored, or it may berelocated to different physicaldisks.

--一些LVM允许在其创建之后改变其特性，甚至在使用时，如resize或mirror或relocated到不同的物理磁盘。

1.4 WhatIs a File System?–什么是文件系统

A filesystem is a data structure built inside a contiguousdisk addressspace. A file manager (FM) is a software package thatmanipulatesfile systems, but it is sometimes called the filesystem. All operating systemshave file managers. The primary taskof a file manager is to allocate anddeallocate disk space intofiles within a file system.

--文件系统是一个连续磁盘空间上的一个数据结构，FM软件可以管理多个文件系统，File manager 有时也称为file system。 所有的操作系统都有file managers。File manager 的主要任务是分配和收集file system 上的磁盘空间。

A file systemallows the disk space to be allocated to a largenumber of files. Each file ismade to appear as a contiguous addressspace to applications such as OracleDatabase. The files may notactually be contiguous within the disk space of thefile system.Files can be created, read, written, resized, and deleted. Eachfilehas a name associated with it that is used to refer to thefile.

--file system 允许磁盘空间分配给大量的文件。每个文件看起来在一个连续的磁盘空间上，但是实际上并不是一个连续的空间。文件可以被创建，读取，resize，和 删除.

A file system iscommonly built on top of a logical volumeconstructed by an LVM. Thus all thefiles in a particular filesystem have the same performance, reliability, andavailabilitycharacteristics inherited from the underlying logical volume. Afilesystem is a single pool of storage that is shared by all the filesin thefile system. If a file system is out of space, then none ofthe files in thatfile system can grow. Space available in one filesystem does not affect spacein another file system. However someLVM/FM combinations allow space to beadded or removed from a filesystem.

An operatingsystem can support multiple file systems. Multiplefile systems are constructedto give different storagecharacteristics to different files as well as todivide theavailable disk space into pools that do not affect each other.

1.5 Benefitsof Using OracleManaged Files

Consider the following benefitsofusing Oracle Managed Files:

--使用OMF有如下好处：

（1）They make the administration of thedatabase easier.

--简化数据库管理

There is no needto invent filenames and define specific storagerequirements. A consistent setof rules is used to name all relevantfiles. The file system defines the characteristicsof the storageand the pool where it is allocated.

（2）They reduce corruption caused byadministrators specifying thewrong file.

--降低管理员指定错误文件的可能性。

Each Oraclemanaged file and filename is unique. Using the samefile in two differentdatabases is a common mistake that can causevery large down times and loss ofcommitted transactions. Using twodifferent names that refer to the same fileis another mistake thatcauses major corruptions.

（3）They reduce wasted disk spaceconsumed by obsolete files.

--减少空间浪费

Oracle Databaseautomatically removes old Oracle Managed Fileswhen they are no longer needed.Much disk space is wasted in largesystems simply because no one is sure if aparticular file is stillrequired. This also simplifies the administrative taskof removingfiles that are no longer required on disk and prevents themistakeof deleting the wrong file.

--DB 会自动移除长时间不用的旧OMF文件。

（4）They simplify creation of test anddevelopment databases.

--简化测试库的创建

You can minimizethe time spent making decisions regarding filestructure and naming, and youhave fewer file management tasks. Youcan focus better on meeting the actualrequirements of your test ordevelopment database.

（5）Oracle Managed Files makedevelopment of portable third-partytools easier.

Oracle ManagedFiles eliminate the need to put operating systemspecific file names in SQLscripts.

1.6 OracleManaged Files andExisting Functionality

Using OracleManaged Files does not eliminate any existingfunctionality. Existing databasesare able to operate as they alwayshave. New files can be created as managedfiles while old ones areadministered in the old way. Thus, a database can havea mixture ofOracle managed and unmanaged files.

二. Enabling the Creation and Useof OracleManaged Files

The followingtable lists the initialization parameters thatenable the use of Oracle ManagedFiles.

启动OMF的初始化参数如下表：

Initialization Parameter	Description
DB_CREATE_FILE_DEST	Defines the location of the default file system directory orOracle ASM disk group where the database creates datafiles ortempfiles when no file specification is given in the createoperation. Also used as the default location for redo log andcontrol filesif DB_CREATE_ONLINE_LOG_DEST_n arenot specified.
DB_CREATE_ONLINE_LOG_DEST_n	Defines the location of the default file system directory orOracle ASM disk group for redo log files and control file creationwhen no file specification is given in the create operation. Bychanging n, you can use thisinitialization parameter multiple times,where n specifies amultiplexed copy of the redo log or control file. You can specifyup to five multiplexed copies.
DB_RECOVERY_FILE_DEST	Defines the location of the Fast Recovery Area, which is thedefault file system directory or Oracle ASM disk group where thedatabase creates RMAN backups when no format option is used,archived logs when no other local destination is configured, andflashback logs. Also used as the default location for redo log andcontrol files or multiplexed copies of redo log and control filesif DB_CREATE_ONLINE_LOG_DEST_n arenot specified.

The file systemdirectories specified by these parameters mustalready exist; the database doesnot create them. The directory mustalso have permissions to allow the databaseto create the files init.

The defaultlocation is used whenever a location is notexplicitly specified for theoperation creating the file. Thedatabase creates the filename, and a file thuscreated is an Oraclemanaged file.

Both of theseinitialization parameters are dynamic, and can beset usingthe ALTERSYSTEM or ALTERSESSION statement.

--这些参数可以使用alter system 或者 altersession 动态修改。

2.1Settingthe DB_CREATE_FILE_DEST InitializationParameter

Includethe DB_CREATE_FILE_DEST initializationparameter in yourinitialization parameter file to identify thedefault location for the databaseserver to create:

-- DB_CREATE_FILE_DEST指定如下参数的默认位置：

（1）Datafiles

（2）Tempfiles

（3）Redo log files

（4）Control files

（5）Block change tracking files

You specify thename of a file system directory that becomes thedefault location for thecreation of the operating system files forthese entities. The followingexamplesets /u01/app/oracle/oradata asthe default directory to usewhen creating Oracle Managed Files:

DB_CREATE_FILE_DEST= '/u01/app/oracle/oradata'

SQL> show parameter db_create_file_dest

NAME                                TYPE       VALUE

----------------------------------------------- -------

db_create_file_dest                 string

--默认为空

2.2Settingthe DB_RECOVERY_FILE_DEST Parameter

Includethe DB_RECOVERY_FILE_DEST and DB_RECOVERY_FILE_DEST_SIZE parametersinyour initialization parameter file to identify the default locationfor theFast Recovery Area.

--这2个参数指定默认FRA的位置，FRA可以包含如下文件：

The Fast Recovery Area contains:

（1）Redo log files or multiplexedcopies of redo log files

（2）Control files or multiplexed copiesof control files

（3）RMAN backups (datafile copies,control file copies, backuppieces, control file autobackups)

（4）Archived logs

（5）Flashback logs

可以通过如下SQL 查看保存的文件类型：

SQL> select file_type fromv$flash_recovery_area_usage;

更多内容参考：

闪回恢复区(Flash Recovery Area)

http://blog.csdn.net/tianlesoftware/article/details/5808279

RMANbackup recovery area 命令

http://blog.csdn.net/tianlesoftware/article/details/5806756

You specify thename of file system directory that becomes thedefault location for creation ofthe operating system files forthese entities. For example:

DB_RECOVERY_FILE_DEST ='/u01/app/oracle/fast_recovery_area'

DB_RECOVERY_FILE_DEST_SIZE = 20G

2.3 Settingthe DB_CREATE_ONLINE_LOG_DEST_nInitializationParameters

Includethe DB_CREATE_ONLINE_LOG_DEST_n initializationparameters in yourinitialization parameter file to identify thedefault locations for thedatabase server to create:

（1）Redo log files

（2）Control files

You specify thename of a file system directory or Oracle ASMdisk group that becomes thedefault location for the creation of thefiles for these entities. You canspecify up to five multiplexedlocations.

--可以指定文件系统目录或者ASM disk group来存放这些文件，最多可以指定5个路径。

For the creationof redo log files and control files only, thisparameter overrides any defaultlocation specifiedinthe DB_CREATE_FILE_DEST and DB_RECOVERY_FILE_DEST initializationparameters.

If you do notspecifya DB_CREATE_FILE_DEST parameter,but you dospecifythe DB_CREATE_ONLINE_LOG_DEST_n parameter,then only redo log filesand control files can be created as OracleManaged Files.

--如果没有指定DB_CREATE_FILE_DEST参数，那么就需要指定DB_CREATE_ONLINE_LOG_DEST_n 参数，并且只有redolog和 control file 可以在OMF 里创建。

It is recommended that you specify at leasttwo parameters. Forexample:

--推荐至少指定2个参数：

DB_CREATE_ONLINE_LOG_DEST_1= '/u02/oradata'

DB_CREATE_ONLINE_LOG_DEST_2= '/u03/oradata'

This allowsmultiplexing, which provides greater fault-tolerancefor the redo log andcontrol file if one of the destinationsfails.

SQL> show parameter db_create_

NAME                                TYPE       VALUE

-----------------------------------------------------------------------------

db_create_file_dest                 string

db_create_online_log_dest_1         string

db_create_online_log_dest_2         string

db_create_online_log_dest_3         string

db_create_online_log_dest_4         string

db_create_online_log_dest_5         string

SQL> show parameter db_recovery

NAME                                TYPE       VALUE

-----------------------------------------------------------------------------

db_recovery_file_dest     string     D:\app\Administrator\flash_recovery_area

db_recovery_file_dest_size          big integer 1300M

三.Creating OracleManaged Files

If you have metany of the following conditions, then OracleDatabase creates Oracle ManagedFiles for you, as appropriate, whenno file specification is given in thecreate operation:

（1）You haveincluded any ofthe DB_CREATE_FILE_DEST, DB_REDOVERY_FILE_DEST,or DB_CREATE_ONLINE_LOG_DEST_n initializationparameters in yourinitialization parameter file.

（2）You haveissued the ALTERSYSTEM statement to dynamically setanyof DB_RECOVERY_FILE_DEST, DB_CREATE_FILE_DEST,or DB_CREATE_ONLINE_LOG_DEST_n initializationparameters

（3）You haveissued the ALTERSESSION statement to dynamically set anyofthe DB_CREATE_FILE_DEST, DB_RECOVERY_FILE_DEST,or DB_CREATE_ONLINE_LOG_DEST_n initializationparameters.

--配置以上参数之后才启用OMF。

If a statementthat creates an Oracle managed file finds an erroror does not complete due tosome failure, then any Oracle ManagedFiles created by the statement areautomatically deleted as part ofthe recovery of the error or failure. However,because of the largenumber of potential errors that can occur with filesystems andstorage subsystems, there can be situations where you mustmanuallyremove the files using operating system commands.

--如果启用了OMF，在创建文件失败后，在OMF recovererror或者failure时时会自动删除这部分创建失败的文件。但是因为一些大量的一些潜在的错误，在一些情况还还是需要手工的从OS上移除这些文件。

The following topics are discussed in thissection:

（1）HowOracleManaged Files Are Named

（2）CreatingOracleManaged Files at Database Creation

（3）CreatingDatafilesfor Tablespaces Using Oracle Managed Files

（4）CreatingTempfilesfor Temporary Tablespaces Using Oracle Managed Files

（5）CreatingControlFiles Using Oracle Managed Files

（6）CreatingRedoLog Files Using Oracle Managed Files

（7）CreatingArchivedLogs Using Oracle Managed Files

关于手工使用createdatabase创建数据库参考：

WindowsOracle 10G手工创建数据库

http://blog.csdn.net/tianlesoftware/article/details/4680230

3.1 How Oracle Managed Files AreNamed

The filenames of Oracle Managed Filescomplywith the Optimal Flexible Architecture (OFA) standard for filenaming.

--Oracle 管理文件的名称通过OFA来实现。 这部分内容参考：

OracleOFA（OptimalFlexible Architecture） 说明

http://blog.csdn.net/tianlesoftware/article/details/7086104

The assigned names are intended to meet thefollowingrequirements:

--命名的相关约束：

（1）Database files are easilydistinguishable from all otherfiles.

（2）Files of one database type areeasily distinguishable fromother database types.

（3）Files are clearly associated withimportant attributesspecific to the file type. For example, a datafile namemay includethe tablespace name to allow for easy association of datafiletotablespace, or an archived log name may include the thread,sequence, andcreation date.

No two OracleManaged Files are given the same name. The namethat is used for creation of anOracle managed file is constructedfrom three sources:

--OMF的文件不会重名，OMF命名由如下三部分组成：

（1）The default creation location

（2）A file name template that is chosenbased on the type of thefile. The template also depends on the operatingsystem platform andwhether or not Oracle Automatic Storage Management is used.

（3）A unique string created by OracleDatabase or the operatingsystem. This ensures that file creation does notdamage an existingfile and that the file cannot be mistaken for some otherfile.

As a specificexample, filenames for Oracle Managed Files havethe following format on aSolaris file system:

destination_prefix/o1_mf_%t_%u_.dbf

where:

--示例中的相关参数说明：

（1）destination_prefix is destination_location/db_unique_name/datafile

where:

1）destination_location is thelocation specifiedin DB_CREATE_FILE_DEST

2）db_unique_name is the globallyunique name(DB_UNIQUE_NAME initialization parameter) of thetargetdatabase. If there isno DB_UNIQUE_NAME parameter,thenthe DB_NAME initializationparameter value is used.

（2）%t is the tablespace name.

（3）%u is an eight-character stringthat guarantees uniqueness

For example, assume the followingparametersettings:

DB_CREATE_FILE_DEST   =/u01/app/oracle/oradata

DB_UNIQUE_NAME = PAYROLL

Then an example datafile name would be:

/u01/app/oracle/oradata/PAYROLL/datafile/o1_mf_tbs1_2ixh90q_.dbf

Names for otherfile types are similar. Names on other platformsare also similar, subject tothe constraints of the naming rules ofthe platform.

The examples onthe following pages use Oracle managed file namesas they might appear with aSolaris file system as an OMFdestination.

Caution:

Do not rename anOracle managed file. The database identifies anOracle managed file based onits name. If you rename the file, thedatabase is no longer able to recognizeit as an Oracle managed fileand will not manage the file accordingly.

--注意，不要对OMF的文件进行重命名，数据库根据OMF的名称来识别文件，如果修改了文件名，那么数据库就会不能识别到这个文件，也就不会管理这个文件。

3.2 CreatingOracle Managed Filesat Database Creation

The actions ofthe CREATEDATABASE statement whenusing Oracle Managed Files are discussed in thissection.

Note:

The rules anddefaults in this section also apply to creating adatabase with DatabaseConfiguration Assistant (DBCA). With DBCA,you use a graphical interface toenable Oracle Managed Files and tospecify file locations that correspond tothe initializationparameters described in this section.

--在DBCA命令中也可以使用OMF，在界面上我们可以选择是否启用OMF，默认是禁用的。

3.2.1 SpecifyingControl Files atDatabase Creation

--创建数据库时指定控制文件位置

At databasecreation, the control file is created in the filesspecifiedbythe CONTROL_FILES initializationparameter.Ifthe CONTROL_FILES parameter isnot set and at least one of theinitialization parameters requiredfor the creation of Oracle Managed Files isset, then an Oraclemanaged control file is created in the default controlfiledestinations. In order of precedence, the default destinationis defined asfollows:

--如果CONTROL_FILES参数没有设置，并且设置了至少一个OMF参数，那么控制文件就会使用OMF目录：

（1）One or morecontrol files as specifiedinthe DB_CREATE_ONLINE_LOG_DEST_n initializationparameter. The file inthe first directory is the primary controlfile.When DB_CREATE_ONLINE_LOG_DEST_n isspecified, the database does notcreate a control filein DB_CREATE_FILE_DEST orin DB_RECOVERY_FILE_DEST (theFast Recovery Area).

（2）If no valueis specifiedfor DB_CREATE_ONLINE_LOG_DEST_n, but values areset forboththe DB_CREATE_FILE_DEST and DB_RECOVERY_FILE_DEST,then thedatabase creates one control file in each location. Thelocationspecifiedin DB_CREATE_FILE_DEST isthe primary control file.

（3）If a value isspecified onlyfor DB_CREATE_FILE_DEST, then the database createsonecontrol file in that location.

（4）If a value isspecified onlyfor DB_RECOVERY_FILE_DEST, then the databasecreates onecontrol file in that location.

Ifthe CONTROL_FILES parameteris not set and none of theseinitialization parameters are set, thenthe Oracle Database default action isoperating system dependent. Atleast one copy of a control file is created inan operating systemdependent default location. Any copies of control filescreated inthis fashion are not Oracle Managed Files, and you mustadda CONTROL_FILES initializationparameter to any initializationparameter file.

If the databasecreates an Oracle managed control file, and ifthere is a server parameterfile, then the database createsa CONTROL_FILES initializationparameterentry in the server parameter file. If there is no serverparameterfile, then you must manually includea CONTROL_FILES initializationparameterentry in the text initialization parameter file.

3.2.2Specifying Redo Log Files at Database Creation

--创建DB时指定redo log 的位置

The LOGFILE clauseis notrequired in the CREATEDATABASE statement, and omitting itprovides asimple means of creating Oracle managed redo log files.Ifthe LOGFILE clause is omitted,then redo log files are created in thedefault redo log filedestinations.

--使用create database 命令创建DB时，logfile参数不是必须的。

In order of precedence, the defaultdestination is defined asfollows:

（1）Ifeitherthe DB_CREATE_ONLINE_LOG_DEST_n isset, then the database creates alog file member in each directoryspecified, up to the valueofthe MAXLOGMEMBERS initializationparameter.

（2）Ifthe DB_CREATE_ONLINE_LOG_DEST_n parameteris not set, butboththe DB_CREATE_FILE_DEST and DB_RECOVERY_FILE_DEST initializationparametersare set, then the database creates one Oracle managed log filememberin each of those locations. The log fileinthe DB_CREATE_FILE_DEST destinationis the first member.

（3）Ifonlythe DB_CREATE_FILE_DEST initializationparameter is specified, thenthe database creates a log file memberin thatlocation.

（4）If onlythe DB_RECOVERY_FILE_DEST initializationparameteris specified, then the database creates a log file member inthatlocation.

The default size of an Oracle managed redolog file is 100MB.

--在OMF 下，默认的redolog是100M。

Optionally, youcan create Oracle managed redo log files, andoverride default attributes, byincludingthe LOGFILE clause but omitting afilename. Redo log filesare created the same way, except for thefollowing: If no filename is providedinthe LOGFILE clauseof CREATE DATABASE, and none of theinitializationparameters required for creating Oracle Managed Files areprovided,then the CREATEDATABASE statement fails.

3.2.3 Specifyingthe SYSTEM andSYSAUX Tablespace Datafiles at Database Creation

--创建DB 时指定SYSTEM 和 SYSAUX 表空间

The DATAFILE or SYSAUX DATAFILE clauseisnot required in the CREATEDATABASE statement, and omitting itprovides asimple means of creating Oracle managed datafilesforthe SYSTEM and SYSAUXtablespaces.

--使用create database 命令时，datafile 参数不是必须的，如果忽略，就会使用OMF。

If the DATAFILE clauseisomitted, then one of the following actions occurs:

（1）If DB_CREATE_FILE_DEST isset,then one Oracle managed datafile forthe SYSTEM tablespace andanotherfor the SYSAUX tablespace arecreatedinthe DB_CREATE_FILE_DEST directory.

（2）If DB_CREATE_FILE_DEST isnotset, then the database createsone SYSTEM andone SYSAUX tablespacedatafile whose names and sizes are operatingsystem dependent.Any SYSTEM or SYSAUX tablespacedatafilecreated in this manner is not an Oracle managed file.

By default,Oracle managed datafiles, including thoseforthe SYSTEM and SYSAUX tablespaces,are 100MB andautoextensible. When autoextension is required, thedatabase extends thedatafile by its existing size or 100 MB,whichever is smaller. You can alsoexplicitly specify theautoextensible unit usingthe NEXT parameterofthe STORAGE clause when youspecify the datafile(ina CREATE or ALTERTABLESPACEoperation).

--默认情况下，OMF 的datafile是100M，并且是autoextensible。在自动宽展时如果没有指定扩展的size，每次扩展也是100M。也可以显示的指定扩展大小。

Optionally, youcan create an Oracle managed datafileforthe SYSTEM or SYSAUX tablespaceand override defaultattributes. This is done by includingthe DATAFILE clause, omittingafilename, but specifying overriding attributes. When a filename isnot suppliedandthe DB_CREATE_FILE_DEST parameteris set, an Oracle manageddatafile forthe SYSTEM or SYSAUX tablespaceis created inthe DB_CREATE_FILE_DESTdirectory withthe specified attributes beingoverridden. However, if a filename isnot suppliedandthe DB_CREATE_FILE_DEST parameteris not set, thenthe CREATEDATABASE statementfails.

--也可以指定datafile参数来覆盖默认的属性。

When overridingthe default attributes of an Oracle managed file,if a SIZE value isspecified butno AUTOEXTEND clause isspecified, then thedatafileis not autoextensible.

--使用覆盖默认属性时，如果只指定了SIZE大小，没有指定autoextent，那么datafile不会autoextensible。

3.2.4Specifyingthe Undo Tablespace Datafile at DatabaseCreation

--创建DB时指定undo datafile

The DATAFILE subclauseofthe UNDO TABLESPACE clause isoptional and a filename is notrequired in the file specification.If a filename is not suppliedandthe DB_CREATE_FILE_DEST parameteris set, then an Oracle manageddatafile is created inthe DB_CREATE_FILE_DEST directory.If DB_CREATE_FILE_DEST isnot set, then the statement fails with asyntax error.

The UNDOTABLESPACE clauseitself is optional inthe CREATEDATABASE statement. Ifit is not supplied, and automatic undo managementmode is enabled(the default), then a default undotablespacenamedSYS_UNDOTS is created and a 20 MBdatafile that is autoextensible isallocated as follows:

--使用create database创建DB时，undotablespace是可选的，如果没有指定，默认的undo表空间名称是SYS_UNDOTS,默认大小是20M，并且默认是autuextensible。

（1）If DB_CREATE_FILE_DEST isset,then an Oracle managed datafile is created in the indicateddirectory.

（2）If DB_CREATE_FILE_DEST isnotset, then the datafile location is operating system specific.

3.2.5Specifyingthe Default Temporary Tablespace Tempfile atDatabaseCreation

--创建DB时指定默认临时表空间

The TEMPFILE subclauseisoptional for the DEFAULT TEMPORARYTABLESPACE clause and afilename is not required inthe file specification. If a filename is notsupplied andthe DB_CREATE_FILE_DEST parameterset, then an Oraclemanaged tempfile is created inthe DB_CREATE_FILE_DEST directory.If DB_CREATE_FILE_DEST isnot set, thenthe CREATEDATABASE statementfails with a syntax error.

The DEFAULTTEMPORARYTABLESPACE clause itself is optional. If it is notspecified,then no default temporary tablespace is created.

The default sizefor an Oracle managed tempfile is 100 MB and thefile is autoextensible with anunlimited maximum size.

--默认OMF下的tempfile是100M，并且是autoextensible，没有最大限制。

3.2.6 CREATEDATABASE StatementUsing Oracle Managed Files: Examples

This sectioncontains examples of the CREATEDATABASE statement whenusing the Oracle Managed Files feature.

3.2.6.1 CREATE DATABASE:Example1

This example creates a database with thefollowing Oracle ManagedFiles:

（1）A SYSTEM tablespacedatafileindirectory /u01/app/oracle/oradata thatis autoextensibleup to an unlimited size.

（2）A SYSAUX tablespacedatafileindirectory /u01/app/oracle/oradata thatis autoextensibleup to an unlimited size. The tablespace is locallymanaged with automaticsegment-space management.

（3）Two online log groups with twomembers of 100 MB each, oneeachin /u02/oradata and /u03/oradata.

（4）If automatic undo management modeis enabled (the default),then an undo tablespace datafileindirectory /u01/app/oracle/oradata thatis 20 MB and autoextensible upto an unlimited size. An undotablespace named SYS_UNDOTS iscreated.

（5）Ifno CONTROL_FILES initializationparameteris specified, then two control files, oneeachin /u02/oradata and /u03/oradata.The controlfilein /u02/oradata is theprimary control file.

The followingparameter settings relating to Oracle ManagedFiles, are included in theinitialization parameter file:

--初始化参数文件配置

DB_CREATE_FILE_DEST ='/u01/app/oracle/oradata'

DB_CREATE_ONLINE_LOG_DEST_1 ='/u02/oradata'

DB_CREATE_ONLINE_LOG_DEST_2 ='/u03/oradata'

The following statementis issued at the SQL prompt:

SQL>CREATE DATABASE sample;

--根据初始化参数创建数据库

To create thedatabase with a locallymanaged SYSTEM tablespace,addthe EXTENT MANAGEMENT LOCAL clause:

SQL>CREATEDATABASE sample EXTENT MANAGEMENT LOCAL;

Without thisclause,the SYSTEM tablespace isdictionary managed. Oraclerecommends that you create a locallymanaged SYSTEM tablespace.

3.2.6.2 CREATE DATABASE:Example2

This example creates a database with thefollowing Oracle ManagedFiles:

（1）A SYSTEM tablespacedatafileindirectory /u01/app/oracle/oradata thatis autoextensibleup to an unlimited size.

（2）A SYSAUX tablespacedatafileindirectory /u01/app/oracle/oradata thatis autoextensibleup to an unlimited size. The tablespace is locallymanaged with automaticsegment-space management.

（3）Two redo log files of 100 MB eachindirectory /u01/app/oracle/oradata. Theyare not multiplexed.

（4）An undo tablespace datafileindirectory /u01/app/oracle/oradata thatis 20 MB and autoextensible upto an unlimited size. An undotablespace named SYS_UNDOTS iscreated.

（5）A controlfilein /u01/app/oracle/oradata.

In this example, it is assumed that:

（1）No DB_CREATE_ONLINE_LOG_DEST_n initializationparametersare specified in the initialization parameter file.

（2）No CONTROL_FILES initializationparameterwas specified in the initialization parameter file.

（3）Automatic undo management mode isenabled.

The following statements are issued at theSQL prompt:

SQL>ALTER SYSTEM SET DB_CREATE_FILE_DEST='/u01/app/oracle/oradata';

SQL>CREATE DATABASE sample2 EXTENTMANAGEMENT LOCAL;

This databaseconfiguration is not recommended for a productiondatabase. The exampleillustrates how a very low-end database orsimple test database can easily becreated. To better protect thisdatabase from failures, at least one morecontrol file should becreated and the redo log should be multiplexed.

3.2.6.3 CREATE DATABASE:Example3

In this example,the file size for the Oracle Managed Files forthe default temporary tablespaceand undo tablespace are specified.A database with the following Oracle ManagedFiles is created:

（1）A 400MB SYSTEM tablespacedatafile indirectory /u01/app/oracle/oradata.Because SIZE is specified,thefile in not autoextensible.

（2）A 200MB SYSAUX tablespacedatafileindirectory /u01/app/oracle/oradata. Because SIZE isspecified,the file in not autoextensible. The tablespace is locallymanagedwith automatic segment-space management.

（3）Two redo log groups with twomembers of 100 MB each, one eachindirectories /u02/oradata and /u03/oradata.

（4）For the default temporarytablespace dflt_ts,a 10 MB tempfileindirectory /u01/app/oracle/oradata.Because SIZE is specified,thefile in not autoextensible.

（5）For the undotablespace undo_ts, a 100 MBdatafile indirectory /u01/app/oracle/oradata.Because SIZE is specified,thefile is not autoextensible.

（6）Ifno CONTROL_FILES initializationparameter was specified, then twocontrol files, one eachindirectories /u02/oradata and /u03/oradata.The controlfilein /u02/oradata is theprimary control file.

The following parameter settings areincluded in theinitialization parameter file:

--初始化参数配置

DB_CREATE_FILE_DEST ='/u01/app/oracle/oradata'

DB_CREATE_ONLINE_LOG_DEST_1 ='/u02/oradata'

DB_CREATE_ONLINE_LOG_DEST_2 ='/u03/oradata'

The following statement is issued at theSQL prompt:

--创建数据库命令

CREATE DATABASE sample3

EXTENT MANAGEMENT LOCAL

DATAFILE SIZE 400M

SYSAUX DATAFILE SIZE 200M

DEFAULT TEMPORARY TABLESPACE dflt_tsTEMPFILE SIZE 10M

UNDO TABLESPACE undo_ts DATAFILE SIZE 100M;

3.3 Creating Datafiles forTablespaces Using Oracle ManagedFiles

--使用OMF来创建表空间

The followingstatements that cancreatedatafiles are relevant to thediscussion in thissection:

（1）CREATE TABLESPACE

（2）CREATE UNDO TABLESPACE

（3）ALTER TABLESPACE ... ADD DATAFILE

When creating atablespace, either a permanent tablespace or anundotablespace,the DATAFILE clause isoptional. When youincludethe DATAFILE clause thefilename is optional.Ifthe DATAFILE clause or filenameis not provided, then the followingrules apply:

（1）Ifthe DB_CREATE_FILE_DEST initializationparameter is specified, thenan Oracle managed datafile is createdin the location specified by theparameter.

（2）Ifthe DB_CREATE_FILE_DEST initializationparameter is not specified,then the statement creating the datafilefails.

When you add adatafile to a tablespace withthe ALTERTABLESPACE...ADDDATAFILE statement the filename isoptional. If the filename is notspecified, then the same rulesapply as discussed in the previous paragraph.

By default, anOracle managed datafile for a permanent tablespaceis 100 MB and isautoextensible with an unlimited maximum size.However, ifinyour DATAFILE clause youoverride these defaults by specifyingaSIZE value(and no AUTOEXTEND clause), thenthedatafileis not autoextensible.

--默认情况，永久表空间默认是100M，并且自动增长，不受限制。可以指定SIZE来覆盖这个默认值。

3.3.1 CREATE TABLESPACE:Examples

The following are some examples of creatingtablespaces withOracle Managed Files.

3.3.1.1 CREATE TABLESPACE:Example1

The followingexample sets the default location for datafilecreationsto /u01/oradata and thencreates a tablespace tbs_1 withadatafile in that location. The datafile is 100 MB and isautoextensible with anunlimited maximum size.

SQL> ALTER SYSTEM SETDB_CREATE_FILE_DEST ='/u01/oradata';

SQL> CREATE TABLESPACE tbs_1;

3.3.1.2 CREATE TABLESPACE: Example2

This examplecreates a tablespacenamed tbs_2 with a datafile inthedirectory /u01/oradata. The datafile initialsize is 400 MB, and becausethe SIZE clause is specified, thedatafile is not autoextensible.

The following parameter setting is includedin the initializationparameter file:

DB_CREATE_FILE_DEST = '/u01/oradata'

The following statement is issued at theSQL prompt:

SQL> CREATE TABLESPACE tbs_2 DATAFILESIZE 400M;

3.3.1.3 CREATETABLESPACE: Example3

This examplecreates a tablespacenamed tbs_3 with anautoextensible datafile inthedirectory /u01/oradata with amaximum size of 800 MB and aninitial size of 100 MB:

The following parameter setting is includedin the initializationparameter file:

DB_CREATE_FILE_DEST = '/u01/oradata'

The following statement is issued at theSQL prompt:

SQL> CREATE TABLESPACE tbs_3 DATAFILEAUTOEXTEND ON MAXSIZE800M;

3.3.1.4 CREATE TABLESPACE:Example4

The followingexample sets the default location for datafilecreationsto /u01/oradata and thencreates atablespacenamed tbs_4 in thatdirectory with two datafiles. Both datafiles havean initial size of200 MB, and because a SIZE valueis specified,they are not autoextensible

SQL> ALTER SYSTEM SETDB_CREATE_FILE_DEST ='/u01/oradata';

SQL> CREATE TABLESPACE tbs_4 DATAFILESIZE 200M, SIZE200M;

3.3.2 CREATE UNDO TABLESPACE:Example

The followingexample creates an undo tablespacenamed undotbs_1 with a datafileinthe directory /u01/oradata. The datafile for theundo tablespace is 100 MBand is autoextensible with an unlimitedmaximum size.

The following parameter setting is includedin the initializationparameter file:

DB_CREATE_FILE_DEST = '/u01/oradata'

The following statement is issued at theSQL prompt:

SQL> CREATE UNDO TABLESPACE undotbs_1;

3.3.3 ALTER TABLESPACE:Example

This exampleadds an Oracle managed autoextensible datafiletothe tbs_1 tablespace. Thedatafile has an initial size of 100 MB anda maximum size of 800MB.

The following parameter setting is includedin the initializationparameter file:

DB_CREATE_FILE_DEST = '/u01/oradata'

The following statement is entered at theSQL prompt:

SQL> ALTER TABLESPACE tbs_1 ADD DATAFILEAUTOEXTEND ON MAXSIZE800M;

3.4 CreatingTempfiles forTemporary Tablespaces Using Oracle Managed Files

The followingstatements that create tempfilesare relevant to the discussion in this section:

（1）CREATE TEMPORARY TABLESPACE

（2）ALTER TABLESPACE ... ADD TEMPFILE

When creating atemporary tablespacethe TEMPFILE clause is optional.If youincludethe TEMPFILE clause, thenthe filename is optional.Ifthe TEMPFILE clause or filenameis not provided, then the followingrules apply:

（1）Ifthe DB_CREATE_FILE_DEST initializationparameter is specified, thenan Oracle managed tempfile is createdin the location specified by theparameter.

（2）Ifthe DB_CREATE_FILE_DEST initializationparameter is not specified,then the statement creating the tempfilefails.

When you add atempfile to a tablespace withthe ALTERTABLESPACE...ADDTEMPFILE statement the filename isoptional. If the filename is notspecified, then the same rulesapply as discussed in the previous paragraph.

When overridingthe default attributes of an Oracle managed file,if a SIZE value isspecified butno AUTOEXTEND clause isspecified, then thedatafileis not autoextensible.

3.4.1 CREATE TEMPORARY TABLESPACE:Example

The followingexample sets the default location for datafilecreationsto /u01/oradata and thencreates atablespacenamed temptbs_1 with atempfile in that location. The tempfile is 100MB and isautoextensible with an unlimited maximum size.

--tempfile默认是100M，自动扩展，不受大小限制

SQL> ALTER SYSTEM SETDB_CREATE_FILE_DEST ='/u01/oradata';

SQL> CREATE TEMPORARY TABLESPACEtemptbs_1;

3.4.2 ALTER TABLESPACE... ADDTEMPFILE: Example

The followingexample sets the default location for datafilecreationsto /u03/oradata and thenadds a tempfile in the default location to atablespacenamed temptbs_1. The tempfile initial size is 100MB. It isautoextensible with an unlimited maximum size.

SQL> ALTER SYSTEM SETDB_CREATE_FILE_DEST ='/u03/oradata';

SQL> ALTER TABLESPACE TBS_1 ADDTEMPFILE;

3.5 Creating Control Files UsingOracle Managed Files

--使用OMF 创建控制文件

When you issuethe CREATECONTROLFILE statement, acontrolfile is created (orreused,if REUSE is specified) inthe files specifiedbythe CONTROL_FILES initializationparameter. Ifthe CONTROL_FILESparameter is notset, then the control file is created inthe default control filedestinations. The default destination is determinedaccording to theprecedence documented in "SpecifyingControlFiles at Database Creation".

If OracleDatabase creates an Oracle managed control file, andthere is a serverparameter file, then the databasecreatesa CONTROL_FILES initializationparameter for the server parameterfile. If there is no serverparameter file, then you mustcreatea CONTROL_FILES initializationparameter manually and include it inthe initialization parameterfile.

If the datafilesin the database are Oracle Managed Files, thenthe database-generated filenamesfor the files must be supplied inthe DATAFILE clause ofthestatement.

If the redo logfiles are Oracle Managed Files,thenthe NORESETLOGS or RESETLOGS keyworddetermines what can besupplied inthe LOGFILE clause:

（1）Ifthe NORESETLOGS keywordis used, then the database-generatedfilenames for the Oraclemanaged redo log files must be suppliedinthe LOGFILE clause.

（2）Ifthe RESETLOGS keyword isused, then the redo log file names can besupplied as withthe CREATE DATABASE statement.See "SpecifyingRedoLog Files at Database Creation".

The sectionsthat follow contain examples of usingtheCREATECONTROLFILE statement withOracle Managed Files.

3.5.1 CREATECONTROLFILE UsingNORESETLOGS Keyword: Example

Thefollowing CREATECONTROLFILE statement is generated byan ALTERDATABASE BACKUP CONTROLFILE TOTRACE statement for a database with Oraclemanageddatafiles and redo log files:

--以下的脚本是使用alter database backupcontrolfile to trace 命令生成的。

CREATE CONTROLFILE

DATABASEsample

LOGFILE

GROUP1 ('/u01/oradata/SAMPLE/onlinelog/o1_mf_1_o220rtt9_.log',

'/u02/oradata/SAMPLE/onlinelog/o1_mf_1_v2o0b2i3_.log')

SIZE 100M,

GROUP2 ('/u01/oradata/SAMPLE/onlinelog/o1_mf_2_p22056iw_.log',

'/u02/oradata/SAMPLE/onlinelog/o1_mf_2_p02rcyg3_.log')

SIZE 100M

NORESETLOGS

DATAFILE'/u01/oradata/SAMPLE/datafile/o1_mf_system_xu34ybm2_.dbf'

SIZE 100M,

'/u01/oradata/SAMPLE/datafile/o1_mf_sysaux_aawbmz51_.dbf'

SIZE 100M,

'/u01/oradata/SAMPLE/datafile/o1_mf_sys_undo_apqbmz51_.dbf'

SIZE 100M

MAXLOGFILES5

MAXLOGHISTORY100

MAXDATAFILES10

MAXINSTANCES2

ARCHIVELOG;

3.5.2 CREATECONTROLFILE UsingRESETLOGS Keyword: Example

The following isan example of a CREATECONTROLFILE statementwiththe RESETLOGS option. Somecombinationof DB_CREATE_FILE_DEST, DB_RECOVERY_FILE_DEST,and DB_CREATE_ONLINE_LOG_DEST_n ormust be set.

CREATE CONTROLFILE

DATABASEsample

RESETLOGS

DATAFILE'/u01/oradata/SAMPLE/datafile/o1_mf_system_aawbmz51_.dbf',

'/u01/oradata/SAMPLE/datafile/o1_mf_sysaux_axybmz51_.dbf',

'/u01/oradata/SAMPLE/datafile/o1_mf_sys_undo_azzbmz51_.dbf'

SIZE100M

MAXLOGFILES5

MAXLOGHISTORY100

MAXDATAFILES10

MAXINSTANCES2

ARCHIVELOG;

Later, you mustissue the ALTER DATABASE OPENRESETLOGS statement to re-create theredo logfiles. This is discussed in "UsingtheALTER DATABASE OPEN RESETLOGS Statement". If the previouslogfiles are Oracle Managed Files, then they are not deleted.

3.6 Creating Redo Log Files UsingOracle Managed Files

Redo log files are created at databasecreationtime. They can also be created when you issue either of thefollowingstatements:

--Online redo log在我们创建DB时会创建，在如下的两条命令下也会创建：

（1）ALTER DATABASE ADD LOGFILE

（2）ALTER DATABASE OPEN RESETLOGS

3.6.1 Usingthe ALTER DATABASE ADDLOGFILE Statement

The ALTER DATABASE ADDLOGFILE statement lets you lateradd a new group toyour current redo log. The filename inthe ADDLOGFILE clause is optionalif you are using Oracle Managed Files. If afilename is notprovided, then a redo log file is created in the default logfiledestination. The default destination is determined according totheprecedence documented in "SpecifyingRedoLog Files at Database Creation".

If a filename isnot provided and you have not provided one ofthe initialization parametersrequired for creating Oracle ManagedFiles, then the statement returns anerror.

The default sizefor an Oracle managed log file is 100 MB.

--默认大小是100M。

You continue toadd and drop redo log file members by specifyingcomplete filenames.

Adding New Redo LogFiles:Example

The followingexample creates a log group with a memberin /u01/oradata and anothermemberin /u02/oradata. The size of each log file is 100MB.

The following parameter settings areincluded in theinitialization parameter file:

DB_CREATE_ONLINE_LOG_DEST_1 ='/u01/oradata'

DB_CREATE_ONLINE_LOG_DEST_2 ='/u02/oradata'

The following statement is issued at theSQL prompt:

SQL> ALTER DATABASE ADD LOGFILE;

3.6.2 Usingthe ALTER DATABASE OPENRESETLOGS Statement

If youpreviously created a control filespecifying RESETLOGS and eitherdidnot specify filenames or specified nonexistent filenames, thenthe databasecreates redo log files for you when you issuethe ALTER DATABASEOPENRESETLOGS statement. The rules for determiningthe directories in which tostore redo log files, when none arespecified in the control file, are the sameas those discussedin "SpecifyingRedoLog Files at Database Creation".

3.7 Creating Archived Logs UsingOracle Managed Files

Archived logs are createdinthe DB_RECOVERY_FILE_DEST locationwhen:

（1）The ARC or LGWR backgroundprocessarchives an online redo log or

（2）An ALTER SYSTEM ARHIVELOGCURRENT statement is issued.

For example,assume that the following parameter settings areincluded in the initializationparameter file:

DB_RECOVERY_FILE_DEST_SIZE = 20G

DB_RECOVERY_FILE_DEST     = '/u01/oradata'

LOG_ARCHIVE_DEST_1        = 'LOCATION=USE_DB_RECOVERY_FILE_DEST'

四. Operation of Oracle ManagedFiles

The filenamesof Oracle Managed Files are accepted in SQLstatementswherever a filename is used to identify an existing file.Thesefilenames, like other filenames, are stored in the controlfile and, if usingRecovery Manager (RMAN) for backup and recovery,in the RMAN catalog. They arevisible in all of the usual fixed anddynamic performance views that areavailable for monitoringdatafiles and tempfiles(forexample, V$DATAFILE or DBA_DATA_FILES).

The following are some examplesofstatements using database-generated filenames:

SQL> ALTER DATABASE

2> RENAME FILE'/u01/oradata/mydb/datafile/o1_mf_tbs01_ziw3bopb_.dbf'

3> TO '/u01/oradata/mydb/tbs0101.dbf';

SQL> ALTER DATABASE

2> DROP LOGFILE'/u01/oradata/mydb/onlinelog/o1_mf_1_wo94n2xi_.log';

SQL> ALTER TABLE emp

2> ALLOCATE EXTENT

3> (DATAFILE'/u01/oradata/mydb/datafile/o1_mf_tbs1_2ixfh90q_.dbf');

You can backupand restore Oracle managed datafiles, tempfiles,and control files as you wouldcorresponding non Oracle ManagedFiles. Using database-generated filenames doesnot impact the use oflogical backup files such as export files. This isparticularlyimportant for tablespace point-in-time recovery (TSPITR)andtransportable tablespace export files.

There are somecases where Oracle Managed Files behavedifferently. These are discussed in thesections that follow.

4.1 DroppingDatafiles andTempfiles

Unlike filesthat are not managed by the database, when an Oraclemanaged datafile ortempfileisdropped, the filename isremoved from the control file and the file isautomatically deletedfrom the file system. The statements that delete OracleManagedFiles when they are dropped are:

（1）DROP TABLESPACE

（2）ALTER DATABASE TEMPFILE ... DROP

--这一点file system 不同，如果使用filesystem，那么我们删除表空间，只是将表空间的新从控制文件中移除，物理上并不会移除对应的物理文件，但是如果使用OMF，删除表空间时，表空间和物理文件会一起删除。

You can also use these statements, whichalways delete files,Oracle managed or not:

（1）ALTER TABLESPACE ... DROP DATAFILE

（2）ALTER TABLESPACE ... DROP TEMPFILE

4.2 Dropping Redo LogFiles

When an Oraclemanaged redo log file is dropped itsOracle ManagedFiles are deleted. You specify the group or membersto be dropped. Thefollowing statements drop and delete redo logfiles:

（1）ALTER DATABASE DROP LOGFILE

（2）ALTER DATABASE DROP LOGFILE MEMBER

4.3 RenamingFiles

The following statements are used to renamefiles:

（1）ALTER DATABASE RENAME FILE

（2）ALTER TABLESPACE ... RENAME DATAFILE

These statementsdo not actually rename the files on theoperating system, but rather, the namesin the control file arechanged. If the old file is an Oracle managed file andit exists,then it is deleted. You must specify each filename usingtheconventions for filenames on your operating system when youissue thisstatement.

--注意，在OMF下的rename并不会实际对物理文件进行rename，而是修改控制文件中记录。

4.4 Managing StandbyDatabases

The datafiles,control files, and redo log files in a standbydatabase can be managed by thedatabase. This is independent ofwhether Oracle Managed Files are used on theprimary database.

When recovery ofa standby database encounters redo for thecreation of a datafile, if thedatafile is an Oracle managed file,then the recovery process creates an emptyfile in the local defaultfile system location. This allows the redo for thenew file to beapplied immediately without any human intervention.

When recovery ofa standby database encounters redo for thedeletion of a tablespace, it deletesany Oracle managed datafiles inthe local file system. Note that this isindependent ofthe INCLUDING DATAFILESoption issued at theprimarydatabase.

五. 小结

在创建instance的时候，我们可以选择存的类型，一般来说，我们都是选择File System。如下图：

一般我们都没有启动OMF，这个可以通过查询我们第二节中提到的相关参数值来确认。如果为空，即没有启动OMF，如：

SQL> show parameter db_create_file_dest;

NAME                                TYPE       VALUE
------------------------------------ -----------------------------------------
db_create_file_dest                 string

在启动OMF之后，我们创建表空间不需要写数据文件路径，名称和大小，默认创建的数据文件大小是100M，自动扩展，unlimited.

并且删除Tablespace时，会连同操作系统级别的物理文件一起删除。

但是实际上对于一个系统，上线之前我们都有详细的规划，所以很少会采用OMF来管理，我们会手动的指定数据文件存放的具体位置，数据文件名称和数据文件的大小，而不是使用OMF，虽然OMF是管理上看起来比较简单，但是实际上管理起来也并见得省多少事。

总之，OMF 是Oracle的一种技术，虽然很少使用它，但是要了解这种技术的存在。