DISK INIT Statement

SQL Server 6.5 supports the use of raw partitions for creating devices with the DISK INIT statement. Raw partitions are disk partitions that have not been formatted with a Windows NT file system, such as FAT and NTFS. In some cases, using devices created on raw partitions can yield a slight performance gain over NTFS or FAT. However, for most installations the preferred method is to use devices on NTFS or FAT partitions.

Syntax

DISK INIT
    NAME = 'logical_name',
    PHYSNAME = 'physical_name',
    VDEVNO = virtual_device_number,
    SIZE = number_of_2K_block
    [, VSTART = virtual_address]

where

NAME

Is the logical name of the database device.

PHYSNAME

Is the drive letter, path, and filename of the database device. To create a device on a raw partition, use only the partition drive letter for the physname parameter in the DISK INIT statement.

VDEVNO

Is the virtual device number.

SIZE

Specifies the size of the database in 2K blocks.

VSTART

Is the starting virtual address.

Example

This example creates a 200 MB device named Inventory_Dev on the raw partition corresponding to the G: drive.

DISK INIT
NAME = 'Inventory_Dev',
PHYSNAME = 'G:',
VDEVNO = 16,
SIZE = 102400
  

Remarks

There are several limitations to consider when using raw partitions:

• Only one device can be created on each raw partition. The logical partition must be configured as a single device, because there is no file system on the raw partition.
• Normal file system operations such as copy, move, and delete cannot be used with raw partitions.
• Devices located on raw partitions cannot be backed up using the Windows NT Backup utility; however, databases or transaction logs stored on raw partitions can still be backed up using the Transact-SQL DUMP statement or through SQL Enterprise Manager.
• File system services such as bad block replacement are not available with raw partitions.