Upon completion of this module, you should be able to :
Create simple, striped, and RAID-5 volumes
Remove a volume
Add a mirror to a volume
Remove a mirror from a volume
Resize a volume(make it larger)
Display properties of a volume
Display volume mapping
Add a file system to a volume
Add a dirty region log to a mirrored volume
Add a log to a volume
Disk Group Review
When designing and administering disk groups, you should be familiar with the following:
Disk groups have two primary functions
The rootdg disk group should be small
Disk groups should have unique names
A disk group should have at least two disks
Moving populated disks requires care
Disk Group Requirements
1,All disk groups across all systems should have unique names
2,All disk groups which reside on one host must have unique names
3,Each system must have a disk group named rootdg
4,In general, the rootdg disk group should be kept small
5,All disk groups must contain at least one disk
Researched Space Allocation
It is frequently better to spend some time analyzing free disk space before creating a volume. Look for patterns of free space that fit your needs. Examples of some commands that can be used to research free space in a disk group are:
# vxdg free
# vxassist maxsize
The simplest form of the command used to create a volume is:
# vxassist make vol02 50m
File system cluster size
You can set the maxconfig parameter for a file system to control the file system I/O cluster size. This parameter specifies the number of 8K blocks that will be culstered together on a write to the disk. The default is 7, which equals 56Kbytes.
To optimize for sequential performance, set maxconfig to(number of spindles in the stripe*the stripe unit size) / the file system block size(8Kbytes). If you are optimizing for random performance, set it to 1.
For best sequential access, the file system cluster size should match some integer multiple of the stripe width as follows:
Four disks in stripe and stripe unit size = 32 Kbytes
maxconfig = 16
Dirty Region Logging
A Dirty Region Log(DRL) keeps a map of data regions that have been modified in mirrored volumes and provides the following advantages:
Highly accelerated mirror resynchronization after a crash
Protection against a second crash during mirror resynchronization
Log Placement
If not properly placed, log volumes can create performance problems. The most important rule about log volumes is:
Log volumes should not reside on the same disk as the volumes they manage
阅读(2343) | 评论(0) | 转发(0) |
