First of all, to tell us what is Quick I / O combined Cached Quick I / O. VERITAS Quick I / O is the oracle database VXFS for a special function. Database and file system integrated with organic at the same time a better help system administrators and database administrators to configure the database server oralce. At the same time, VERITAS Quick I / O is also for the creation of common documents to provide the bare device interface. It must avoid the overhead of file system operation (such as: data lock level operation) to provide kernel asynchronous I / 0 (KAIO), therefore, Quick I / O for file system management and performance of equipment with bare to expediency. Cached Quick I / O is an enhanced Quick I / O, it can help better use of oracle DBA memory efficiency of large-scale systems.
     In other words, the traditional file system, as a result of the document the existence of single write lock, leading to the file system read and write in the complicated environment of relatively low efficiency. Bare equipment and the problem does not exist. Quick I / O is actually the elimination of a single write lock on the file system to improve efficiency.
     Cached Quick I / O is the Quick I / O of an enhanced deal with the ideological core of the record of the visit is to continue to document the frequency, the document will be the hottest on the cache, the cache hit rate to increase, thereby enhancing the document efficiency of the system.
Use Cached Quick I / O can be better run in the file system database.
System to facilitate the provision of the document production environment I / O statistics.
    Collection of documents in the file I / O statistics.
Use Cached Quick I / O to facilitate analysis of the I / O statistics.
Do not use Cached Quick I / O in a larger percentage difference.
A simple understanding of these two characteristics:
1) Quick I / O address is the traditional unix file system the issue of single write lock.
2) Cached Quick I / O address is the cache hit rate.

First to introduce the Quick I / O.
    First of all, to confirm that the file system is vxfs. This is characteristic of the VERITAS storage foundation.
    Need to buy veritas license
    Veritas provided by the use of scripts to create the data files needed to be translated into a list of
  $ / opt / VRTSdbed / bin / qio_getdbfiles the implementation of the script may be needed to be translated into all vxfs data files are output to mkqio.dat file. Script directory in / opt / VRTSdbed / bin. Version of the old name of this script: getdbfiles.sh.
     This script will automatically rule out all types of data files for temproary. After the conversion of these documents, oracle will not be able to write.
  mkqio.dat documents can also create their own hand. However, we must pay attention to rule out temporary datafiles. File contains data files and the size of sample is as follows:

$ cat mkqio.dat
/ db01/file1 210358
/ db01/file2 157996
/ db01/file3 38098
/ db01/file4 394935
/ db01/file5 911787
/ db01/redo_t04.log 209715
In the above process, the need to open the database.

  $ / opt / VRTSdbed / bin / qio_convertdbfiles, the implementation of the script into mkqio.dat data files in the list. Version of the old name of this script: mkqio.sh
Before the conversion, you must first close the database.
These scripts are oracle user.
Let me now how to enable Cached Quick I / O do? Vxtunefs can order through the use of qio_cache_enable signs order to set to 1. Command as follows:

# Vxtunefs-s-o qio_cache_enable = 1 / mount_point

For example:

# / Opt / VRTSvxfs / sbin / vxtunefs-s-o qio_cache_enable = 1 / happy

vxfs vxtunefs: Parameters successfully set for / happy

Then see how close the Cached Quick I / O do? Vxtunefs can order through the use of the command will be changed to 0 to qio_cache_enable logo set. Command as follows:

# Vxtunefs-s-o qio_cache_enable = 0 / mount_point

For example:

# / Opt / VRTSvxfs / sbin / vxtunefs-s-o qio_cache_enable = 0 / happy1

vxfs vxtunefs: Parameters successfully set for / happy1

If you want Cached Quick I / O settings without restarting the impact of fixed, on the need to / etc / vx / tunefstab file the following entry qio_cache_enable increase.
If that does not exist tunefstab words, one needs to add a hand on it.
For example:
/ dev / vx / dsk / datadg / datavol qio_cache_enable = 1
/ dev / vx / dsk / datadg / myvol qio_cache_enable = 1

Can vxtunefs command to view the Cached Quick I / O situation.
Command as follows:
# Vxtunefs / mount_point

For example:
# / Opt / VRTSvxfs / sbin / vxtunefs / happy
Filesystem i / o parameters for / happy
read_pref_io = 65536
read_nstream = 1
read_unit_io = 65536
write_pref_io = 65536
write_nstream = 1
write_unit_io = 65536
pref_strength = 10
buf_breakup_size = 131072
discovered_direct_iosz = 262144
max_direct_iosz = 1048576
default_indir_size = 8192
qio_cache_enable = 1
write_throttle = 15104
max_diskq = 1048576
initial_extent_size = 8
max_seqio_extent_size = 2048
max_buf_data_size = 8192
hsm_write_prealloc = 0

# / Opt / VRTSvxfs / sbin / vxtunefs / happy1
Filesystem i / o parameters for / happy1
read_pref_io = 65536
read_nstream = 1
read_unit_io = 65536
write_pref_io = 65536
write_nstream = 1
write_unit_io = 65536
pref_strength = 10
buf_breakup_size = 131072
discovered_direct_iosz = 262144
max_direct_iosz = 1048576
default_indir_size = 8192
qio_cache_enable = 0
write_throttle = 15104
max_diskq = 1048576
initial_extent_size = 8
max_seqio_extent_size = 2048
max_buf_data_size = 8192
hsm_write_prealloc = 0

At the same time, individual documents can also set up Cached Quick I / O. Qioadmin order to set up the use of documents Cached Quick I / O. Qiostat or other use of analytical tools to analyze the documents Cached Quick I / O. When individuals file to open and close the Cached Quick I / O requires the database is running. Qiostat through the use of document order to observe whether the cache hit rate sufficient to open or closed to change the Cached Quick I / O for individual files also need to consider setting up cache.
Close personal documents Cached Quick I / O command set used qioadmin flag OFF.
Command as follows:
# Qioadmin-S file = OFF mount_point
For example:
# / Opt / VRTSvxfs / sbin / qioadmin-S myfile = OFF / happy

Open the personal files of Cached Quick I / O command set used qioadmin flag ON.
Command as follows:
# Qioadmin-S file = ON mount_point
For example:
# / Opt / VRTSvxfs / sbin / qioadmin-S myfile = ON / happy

If you want to set up has been the entry into force of the impact from the need to restart the file / etc / vx / qioadmin entries corresponding increase.
If this file does not exist, you can manually add up.
recommend the use of cache to store qioadmin need to add attributes Inode. These settings can be re-mount and system restart has been the entry into force, but these attributes through general backup methods can not be backed up. They can not be restored, so that any restoration of a file system needs to re-set attributes. If the / etc / vx / qioadmin increase of Quick I / O file system entries, these do not need to do work.

/ etc / vx / qioadmin the contents of the paper is as follows:
Happy in the file system as an example:
device = / dev / vx / dsk / datadg / datavol
myfile, OFF
mydata, ON

Qioadmin can use the-P option command to view the document now Cached Quick I / O situation.
Command as follows:
# Qioadmin-P file mount_point

For example:

# / Opt / VRTSvxfs / sbin / qioadmin-P myfile / happy

myfile, OFF

# / Opt / VRTSvxfs / sbin / qioadmin-P mydata / happy

mydata, ON