To SQLSERVER energy high efficiency operation, SQLSERVER disk subsystem is an important aspect
Avg. Disk Sec / Read this counter is an average per second of data read from disk
The following list shows the range of the counter values, and indicates the range in which the counter means
Less than 10 ms - very good
at 10 - 20 ms between - also
at 20 - 50 ms between - slow, need to focus on
greater than 50 ms - severe I / O bottlenecks
Disk Performance Test Tool
（1）CrystalDiskMark
( 2 ) HDTune hard disk check and repair tool
（3）ATTO Disk Benchmark
Disk drive location :
For different purposes, you need to use a different drive to store the following things
separate disk latency requirements:
a database of small to 15ms
The transaction log small to 2ms
Tempdb database small to 2ms
This means that, Tempdb placed on a separate physical disk, transaction log files on separate physical disks, data files on a separate physical disk, the operating system on a separate physical disk,
Database backup files on separate physical disks
Our general approach: there can be so many separate physical disk, generally is made of the storage disk array
C disk to put the operating system files
D disk release data files and transaction log files and Tempdb data files and Tempdb log file
E disk database backup files put
Of course, if there is sufficient disk arrays and,
Recommended operating system files , mechanical rotary raid1
Data files , mechanical rotary raid10
Tempdb data files and Tempdb log file to put SSD raid1
The transaction log file to put SSD raid1
When the disk formatting time, to be stored SQLSERVER disk data files and log files, try not to use the default disk allocation unit
Use 64k cluster size Allocation Unit to format the disk