In Oracle, four file system mechanisms can be used to store data. (A database may contain all the following types of files)
· "Cooked" operating system (OS) File System
These files are stored in the OS file system. You can use the OS tool to move files. The Cooked file system usually caches information for you when you read or write a disk.
· Raw partition (also called raw partition)
This is not a file, but an original disk. You cannot use the OS tool to view its content. These are some of the large sectors on the disk, without any file systems. For Oracle, the entire original partition is a large file. Unlike the cooked file system, the cooked file system may have dozens or even hundreds of database data files. Currently, only a few Oracle installations use the original partition, because the management overhead of the original partition is large. The original partition is not a buffer device, and all I/O operations are directly I/O, without any OS buffer for data (however, this is usually an advantage for databases ).
· Automatic Storage Management (ASM)
This is a new feature of Oracle 10g Release 1 (both the Standard Edition and Enterprise Edition provide this feature ). ASM is a file system specially designed for databases. You can simply regard it as a database file system.
· Cluster File System
This file system is dedicated to the RAC (cluster) Environment and looks like a cooked file system shared by multiple nodes (computers) in the cluster environment. Oracle Cluster File System (OCFS) is a new File System launched by Oracle in this field. It can only be used on Windows and Linux. Other third-party developers also provide some certified cluster file systems, which can also be used in Oracle. The cluster file system extends the advantages of the cooked File System to the cluster environment.
The Oracle storage system is summarized as follows:
(1) A database consists of one or more table spaces.
(2) A tablespace consists of one or more data files. These files can be cooked files in the file system, original partitions, database files managed by ASM, or files on the cluster file system. Table space Segment (Segment ).
(3) segments (TABLE, INDEX, etc.) are composed of one or more segments (Extent. Segments are in the tablespace, but can contain data in multiple data files in the tablespace.
(4) The partition is a logical continuous Block on the disk ). The partition is only in one tablespace and is always in one file in the tablespace.
(5) blocks are the smallest allocation units in the database and the smallest I/O units used by the database.
Generally, Oracle reads and writes data from a disk as a block.
A database can contain multiple block sizes, so that you can use a table space that can be transferred in more cases. If the tablespace can be transferred, DBA can move or copy formatted data files from one database and put them in another database.
The DATABASE also has a default block size, that is, the size specified in the initialization file when the create database Command is executed. The SYSTEM tablespace always uses the default block size.
You can create other tablespaces based on non-default block sizes. Within all given tablespaces, the block size is consistent.
From: NowOrNever