Dynamic Disks 101
Written by the Stoic Joker
Windows XP Home does not support Dynamic Disks.
Windows 2000/XP Professional does not support Dynamic Disks when installed on a laptop.
Only Windows 2000/2003 Server Operating Systems support Fault Tolerant Dynamic
Lets start by defining what a dynamic disks
is; A dynamic disk is a disk that has been initialization with the dynamic
storage standard which is a single partition that includes the entire disk that
can be divided into volumes. These volumes can consist of a portion, or
portions, of one or more physical disks. They can also be created, resized,
and formatted, without requiring that the operating system be rebooted before
using them. There are five different types of volumes that can
be created on a dynamic disk.
|Contain disk space from a single disk
and are not fault tolerant
Available in Windows 2000/2003 Server
Available in Windows 2000/XP professional
|Include disk space from multiple disks
(up to 32) into one logical volume. Data is written to the first
disk until it is completely filed then to the 2nd, 3rd, etc. If any
disk in a spanned volume fails, the data in the entire volume is
lost. Spanned volumes are not fault tolerant.
Stripped (RAID-0) Volumes
|Include disk space from multiple hard
disks (up to 32) into one logical volume. In a striped volume data
is written to all disks at the same rate which increases the volumes
I/O performance assuming that the disks are either SCSI or on
separate controllers. If any disk in a stripped volume fails, the
data in the entire volume is lost. RAID-0 volumes are not fault
Mirrored (RAID-1) Volumes
|Consist of two identical copies of a
simple volume, each on a separate hard disk. If either disk in a
mirrored volume fails, the data is still intact on the second
"mirrored" disk. Mirrored volumes are fault tolerant
Available in Windows 2000/2003 Server
|Are a fault-tolerant version of the
striped volume. A parity-information stripe is added to each disk in
the volume. If any disk in a RAID-5 volume fails the
parity-information stripe is used to reconstruct the data. A minimum
of three hard disks are required to create a RAID-5 volume.
|Note: As I have no intentions of debating the pros,
cons, & capabilities of various third party utilities here...The following rules pertain to
the options that are available in the MS Windows Disk Management
Simple Volumes You can create a simple volume and
format it with FAT, FAT32, or NTFS. But... Simple volumes can only be extended or spanned only if they
are formatted with the NTFS file system.
Spanned Volumes can be used to combine areas of unused
disk space of various sizes into one logical volume that can be accessed under a
single drive letter. You can create a spanned volume and format it with FAT,
FAT32, or NTFS. But... Spanned volumes can only be extended to include new areas
of free space after they're created if they
are formatted with the NTFS file system. You can also be extended after being created. You can not however remove a physical disk
from a spanned volume, or a physical disk with space used by a spanned volume
without backing up the data (or kissing it good bye), deleting and then
recreating the spanned volume (e.g. You can make it bigger, but you can't make
Stripped Volumes offer the best performance of all the
disk management strategies. In a striped volume, data is written evenly across
all physical disks in 64-KB units, and all of the physical disks that belong to
the striped volume perform as a single physical disk. This allows the OS to
process concurrent I/O commands simultaneously on all hard disks which is why
striped volumes can increase the speed of system I/O, assuming that the hardware
is configured in a fashion that will allow it to do so (i.e. Using two physical
disks to create a stripped set will not increase I/O performance if they are
attached to the same IDE controller). Stripped volumes require that the free
space used to create them be contiguous and the same size on all drives used.
Stripped volumes can not be extended/resized.
Mirrored Volumes create a mirror image of drive 'A' on
drive 'B' in real time, so if a "fault" occurs (e.g. drive 'A' dies) all of the
data is still intact and available on drive 'B', hence the term fault tolerant.
While the System and Boot partitions (e.g. Drive C:) can be mirrored
using dynamic disk it's really not a good idea as it tends to degrade performance severely by trying to mirror the
Hardware mirroring which is much faster is still not a good idea unless your
using SCSI drives. Critical
databases, files, etc. should never be stored on drive C: for a multitude of
reasons, but that is beyond the scope of this tutorial. Mirrored volumes require
that the free space used to create them be contiguous and the same size on both
RAID-5 Volumes being a stripped volume with parity
information are the faster of the two fault tolerant disk configurations
available. If a "fault" occurs (e.g. one of the drives die), the data contained
on the dead drive can be regenerated (recreated) using the parity information
contained in the other disks in the RAID-5 volume. This however is not an excuse
for not having a good backup strategy as it is not impossible for two drives to
die at the same time, or one of the ever popular "latest & greatest" viruses
could just eat the whole shebang. RAID-5 volumes require that the free space
used to create them be contiguous and the same size on all drives used. RAID-5
volumes can not be extended/resized.
The conversion from basic to dynamic
storage requires a minimum of 1MB of un-partitioned disk space at the end of the
drive where the dynamic disk's volume information is stored. When creating
partitions with the Windows setup utility, you can not create a partition or
group of partitions that use the maximum size of the drive. This is due to the
setup utility automatically reserving 1MB or one cylinder whichever is greater
at the end of the drive which allows you to upgrade the drive dynamic disk at a
Any basic partitions that were on the
drive before it was upgraded to dynamic disk can not be extended,
spanned, or stripped, they can however be mirrored. Windows 2000 could not be
directly installed on a dynamic volume because the partition tables used by
dynamic disks are not stored in the location that an Intel-based computer's BIOS
expects to find them. This issue was later addressed in Windows XP which can be
installed directly on a dynamic volume after the master boot record (MBR)
partition table information is created using the Diskpart utility's retain
command, see links below for details.
Switching Between Basic & Dynamic Storage
You can upgrade a
disk from basic
storage to dynamic storage at any time without loss of data. Reverting from
dynamic storage back to basic storage requires the removal (e.g. deleting...) of
all volumes that are on the dynamic disk before the change will take place.
Upgrading a basic disk that contains an extended partition with logical drives
is rather pointless and tends to create a huge mess.
|Note: Folder Mount Points; The
practice of attaching a partition to an empty folder so that it can
be accessed without requiring a drive letter, does not
require that either disk be dynamic. It requires only that the
partition the folder resides on use the NTFS file system, and that
the folder in question must be empty.
The System and Boot Volumes (/Partitions)
The distinction between these two has been the cause for much confusion due to
the fact that they are both typically drive C:. Back in the days when
WindowsNT 4.0 and NTFS were new, it was a common practice to use a small Fat16
partition for the system files (NTLDR, NTDETECT.COM, BOOT.INI, & etc.) and a
copy of DOS which listed as a boot option and would allow limited access to the
computed if the primary (NT) OS failed to boot. That was and still is known as
the system partition although the copy of DOS is now optional.
The booted Operating System's files (the
WinNT or Windows folder, Program Files, & etc.) were then installed to a
separate NTFS partition to take advantage of the NTFS file and folder security
options. That was and still is know as the boot partition.
This practice was necessary to simplify
recovery from a system failure because there was no recovery console available
at that time.
Dual Booting with Dynamic Disks
Never upgrade a dual boot machine to dynamic
disk if both operating systems are on the same physical disk, dynamic disks can
only be "owned" (e.g. accessed on local machine) by one OS, so the second OS on
the disk will become unbootable (totally hosed). If dual booting with Win 9x or
NT4 they will not be able to read the data contained on a dynamic disk. Also
while any of the Win2000 or newer MS operating systems can access dynamic
storage drives the resulting conflict over which OS "owns" the dynamic disk can
end in the loss of all the data in contains. Any operation system be it DOS,
Win9x/NT4, Mac, Linux, etc. can however access the data contained on a dynamic
disk if it is done across the network.
Additional Reading and