Free spacetable (FST)
FST indicates the free space contained in the ATB. ASM generates a disk FST when the disk is selected for assignment. This allows ASM to skip the space occupied by the ATBs. FST contains entries for each ATB. Each entry describes the total number of extents size and the idle extents. The purpose of FST is to optimize allocation operations. In fact FST is located in Allocationtable. The FST is located at the second block (Block 1) of each at.
Partnershipstatus Table (PST)
PST Tracks disk group eligibility and inter-disk mate relationships. ASM generates a PST to determine whether a set of disk-mounted disks using a group of online disks meets the needs. Each disk in a disk group retains the second au (au 1) for PST use. However, at any one time, a disk group has only a certain number of active PSTs. The number of PST depends on the disk group redundancy and the number of failed groups (Failure group). The external redundant disk group has a PST. If there are at least 3 failure groups, the Normal-redundancy disk group contains 3 PSTs. However, each failure group of the Normal-redundancy Disk group has a PST. The High-redundancy disk group has 5 PST if there are enough failure groups present. However, each defunct group in the High-redundancy Disk group has a PST.
PST is a shadow paging used to ensure that the active PST on disk is always consistent, in time PST across multiple blocks. AU 1 satisfies the need to write two parts of PST. One is an active copy, and the other copies are used for PST updates. Once the PST is updated, the previous shadow becomes an active PST, and the other copy becomes a shadow for the next update to be used. The RDBMS management control file is a similar technique.
PST The first part of the information is about PST own:
N versionnumber
N Timestamp
n pstsize (number of disks)
N numberof PST copies
n listof disk containing the PST
N Valueof compatible.asm (if compatible. asm>=11.1)
ASM stores stored compatible.asm in PST to reduce the amount of storage, and this information modification needs to be submitted to the member for modification. True as described in this chapter of the disk header, ASM also updates the COMPATIBLE.ASM value of all disk headers, preventing 10g software products from mounting incompatible disk groups, while Oracle DATABASE 11g uses the values in PAT.
PST contains a set of entries that describe each disk in a group of disks. For each disk, PST tracks the following:
n Disk status (online or offline, for example)
N Number of Members
N Member List
The last block of the PST saves the heartbeat of the disk group. ASM uses this block to ensure that disk groups are not mounted simultaneously in different clusters. Just like rac,asm under the protection of cluster synchronization Service (CSS) in order to avoid the brain fissure in the cluster. PST Heartbeat is an additional mechanism that blocks brain fissures on the cluster. This is like a heartbeat when the RDBMS writes control files avoid two instances of the same cluster mount a database at the same time.
Typically, a disk group must have access to a certain number of PST to mount a disk group. Normal-redundancy Disk group requires 2 failed disk groups high-redundancy The disk group requires a degradation of 5 failed disk groups and requires fewer PST to mount the disk group.
Virtual meta data
The virtual metadata is stored in an ASM file. A directory is a source data file that an ASM instance accesses by exclusive means. The catalog file number starts at 1. Not only do RDBMS instances register to save ASM files, but ASM instances are also available. The registration number is under 2,551. The file number is not mentioned in this section to preserve the future use. V$asm_file does not display metadata catalogs or registrations.
Like other ASM files, virtual metadata files are mirrored by the disk group redundancy type. ASM does not provide mirroring protection for external redundancy disk groups. Virtual metadata provides 3 images in the normal-and high-redundancy disk groups. The following data structure forms the virtual metadata:
N File Directory
N Disk Directory
n Active Change Directory (ACD)
N Continuing Operations Directory (DOD)
N Template Directory
N Alias Directory
N Attribute Directory
N staleness Directory
N staleness Registry
File Directory
File directory contains all the metadata related to the ASM file. The directory contains entries for each ASM file and these entries are indexed using numeric values. The file number of the file directory in each disk group is 1 (F1).
File directory contains the following fields:
N Incarnation number
n File Size
N File Block Size
n File Type
N Redundancy (None,normal,or high)
n Striping (coarse or fine-grained)
n File creation time
N File modification time (when the file was last opened or written)
N File Layout
ASM If a file is deleted and then the file number of the file is reused to create a new file, the file's avatar number is unique and not the same as before. The avatar is generated based on the timestamp when the data file was created, ensuring that the ASM file name is unique when the avatar is the same.
The block size of the ASM file does not depend on the ASM metadata block size. The 4k block size is used for all ASM meta-directories. The RDBMS data file can be set to 2k,4k,8k,16k, or 32K in the CREATE table space. The redo log file for an RDBMS is generally set to 512byte block size. The ASM and trace logical File block sizes (as seen in the data block) also track the occupancy of the physical space in the disk group and are placed in the audit file for future reference. The RDBMS also provides file types when files are created. For files that are not explicitly created by an RDBMS instance, such as those created with the XMLDB FTP command or the asmcmd CP command, ASM checks the file header to determine the file type when creating the file.
The meaning of create time is obvious. However, the filemodification time is not updated every times the data file is written. Instead, file modification time updates the timestamp when the files need to be written and open. This means that modification time is modified to not write to the data, stored modification may be earlier than the last modification of the file content. Similarly, to reduce contention in the buffer cache for file directory blocks in an ASM instance, Modificationtime is truncated on time per hour. Therefore, if several instances in the cluster open a file in a very short time, only the first instance accesses the file and updates the files modification.
The file distribution information consists of a series of extentpointers. Extent pointer specifies the disk number and AU number where the extent is located. The FileDirectory entry contains the first 60 extentpointer of each file, sometimes referred to as direct extent (directly extended). FileDirectory the rest of the space to save indirect extent (non-direct extension). Indirect extent is the other virtual metadata extent includes the ASM extent pointer. Each of the indirect extent is an AU. Each file directory entry can be stored in 300 indirect extent.
The concept of indirect pointer is found in traditional file systems, such as the UNIX BSD file system. The file layout of the push show is also seen as the extent map, and the details of how striping and redundancy affect the layout of the file will be discussed in Chapter 11th, "ASM Operations".
Figure 10-1 shows a file large enough to hold the indirect extent. Note Indirect extent is metadata that extends beyond file directory. The disk group in the example has 10 ASM disks (D0 through D9).
The relationship between Extentpointer and Allocat table is shown in Figure 10-2. In this example, the file 256 has 5 extent (0 to 4) and the file 257 has 4 extent (0 to 3). The disk group has 3 disks (0 to 2). The diagram shows only the files involved in the example, which is a small part of the allocation table.
--------------------------------------------------------------------------------------------------------------- --
This site is marked original and translated are original articles, the article allows reprint, but must be linked to the source address,
otherwise be held responsible for legal ! there are inevitably omissions in the article welcome netizens to criticize correct.
qq:173386747
Email: [Email protected]
blog:http://blog.csdn.net/card_2005