Implementing RAID on Windows 2000 Servers

Rotational latency (the time for the disk to rotate to the requested sector from the sector it was at when the head reaced the correct cylinder), Is minimized in for the next sector since once the heads arrived on-cylinder, subsequent sectors ( not necessarily requested [yet] in the track were read to the onboard cache. With 2 drives acting as one the cache is doubled!

This assumes that the less expensive multiple disks are of the same performance family as the anticipated large disk.
Upgrading to a higher performance family, for example 10k (10,000 RPM), may reduce the transfer rate. as much may be 50% faster then 75,00RPM disks.
The specifications of the particular disk under consideration must be examined recording density contributes to the amount of data stored per track.
Track-to-track and maximum seek time may be significantly shorter in higher performance disks. Although more expensive, higher performance disks are usually built with higher expected duty cycle and larger mean time between failure.

striping decreases reliability since a failure of either disk causes all data to be unavailable.

Other considerations include the availability of controller channels and there is an increasing concern regarding power and related cooling requirements.

Implementing RAID on Windows 2000 Servers^†

TechNet Home > Products & Technologies> Server Operating Systems >

from Chapter 11, Microsoft Windows 2000 Administrator's Pocket Consultant by William R. Stanek.

Windows 2000 server supports software implemented disk mirroring, disk striping,
and disk striping with parity. Some operating systems, don't support RAID.
If you dual boot to one of these operating systems, RAID-configured drives will be unusable.

Implementing RAID 0: Disk Striping
Two or more volumes—on a separate disk—are configured as a stripe set. Data written to the stripe set is broken into blocks that are called stripes. These stripes are written sequentially to all drives in the stripe set. You can place volumes for a stripe set on up to 32 drives, but in most circumstances sets with two to five volumes offer the best performance improvements over non-RAID volumes because data from multiple drives is simultaneously accessed. Beyond this, the performance improvement decreases significantly.
The advantage of disk striping is speed, not reliability. Data can be accessed on multiple disks using multiple disk heads, which improves performance considerably. If any disk in the stripe set fails, all data in the stripe set is lost. You'll need to recreate the stripe set and restore the data from backups. Data backup and recovery are discussed in Chapter 14.
The boot and system volumes can't be part of a striped set.
When you create stripe sets, use volumes that are approximately the same size. Disk Management bases the overall size of the stripe set on the smallest volume size. Specifically, the maximum size of the stripe set is a multiple of the smallest volume size. For example, if you have three physical disks and if the smallest volume is 50 MB, the maximum size for the stripe set is 150 MB.
To maximize performance of the stripe set, there are several things you can do:
•
Use disks that are on separate disk controllers. This allows the system to simultaneously access the drives.
•
Don't use the disks containing the stripe set for other purposes. This allows each disk to dedicate its time to the stripe set.

Create a stripe:
1.
In the Disk Management Graphical view, right-click an area marked Unallocated on a dynamic disk and then choose Create Volume. This starts the Create Volume Wizard. Read the welcome dialog box, and then click Next.
2.
Select Striped Volume as the volume type. Create the volume as described previously in this chapter in the section entitled "Creating Volumes and Volume Sets." The key difference is that you need at least two dynamic disks to create a striped volume.
3.
You can't expand a stripe set.

Implementing RAID 1: Disk Mirroring
Identically sized volumes on two different disks create a redundant data set. The disks are written with identical sets of information, and if one of the disk fails, data is be obtained from the other disk.
Disk mirroring offers the same fault tolerance as disk striping with parity. Because mirrored disks don't need to write parity information, they offer better write performance. Disk striping with parity offers better read performance because read operations are spread over multiple drives.
The major drawback to disk mirroring is that it cuts the usable storage space in half.
Note: Unlike with disk striping, you can mirror any volume including the boot and system volumes . The mirrored disks may be on separate controllers. This provides increased protection against failure of the controller. If one of the controllers fails, the disk on other controller is still available. Using two separate controllers is disk duplexing. Figure 11-5 shows the difference between the two techniques. Disk mirroring typically uses a single drive controller, but disk duplexing uses two drive controllers. If one of the mirrored drives in a set fails, disk operations can continue. Here, when users read and write data, the data is written to the remaining disk. You'll need to break the mirror before you can fix the mirror. To learn how, see the section of this chapter entitled Managing RAIDs and Recovering from Failures."
Creating a Mirror Set in Disk Management

In the Disk Management , right-click an area marked Unallocated on a dynamic disk, and then choose Create Volume. This starts the Create Volume Wizard. Read the welcome dialog box, and then click Next.
Select Mirrored Volume as the volume type. Create the volume as described previously in the section of this chapter entitled "Creating Volumes and Volume Sets." The difference is that you must create two identically sized volumes and these volumes must be on separate dynamic drives.
Figure 11-5: While disk mirroring typically uses a single drive controller to create a redundant data set, disk duplexing uses two drive controllers.

Mirroring is transparent to users. Users see the mirrored set as a single drive that they can access and use like any other drive.

Note: The status of a normal mirror is Healthy.
During the creation of a mirror, you may see a status of Initializing.
This tells you that Disk Management is setting up the mirror.

Mirroring an Existing Volume

The volume to mirror must be a simple volume(i.e. not extended[?]),
and you must have an area of unallocated space on a second dynamic drive.

In Disk Management
1.
Right-click the simple volume you want to mirror, select Add Mirror.
2.
Use the wizard dialog boxes to select and configure the second volume.

Implementing RAID 5: Disk Striping with Parity A minimum of 3 disks are required. The volume on these disks are sized identically by Disk Management. Although you can place volume for a stripe set on up to 32 disks, in most circumstances sets with two to five volumes offer the best performance improvements. Beyond this, the performance improvement decreases significantly.
RAID 5 is an enhanced version of RAID 1 (striping) with the addition of fault tolerance which ensures that the failure of a single disk won't bring down the set. Instead, the set continues to function with disk operations directed at the remaining disks in the set.
RAID 5 writes parity checksums with the blocks of data. If any of the drives in the stripe set fails, the parity information is used to recover the data. (This process, called regenerating the striped set, is covered in the section of this chapter entitled "Managing RAIDs and Recovering from Failures.") If two disks fail the parity information isn't sufficient to recover the data, and you'll need to rebuild the striped set from backup.
Note: The boot and system volumes can't be part of a striped set. Creating a Stripe Set with Parity

In the Disk Management, right-click an area marked Unallocated on a dynamic disk and then choose Create Volume.
Select RAID-5 Volume as the volume type. Select free space on three separate dynamic drives
You can't expand a stripe set once it's created.

Managing RAIDs and Recovering from Failures

Managing mirrored drives and stripe sets is somewhat different from managing other drive volumes,
especially when it comes to recovering from failure.

Breaking a Mirrored Set

You may want to break a mirror for two reasons:
•
If one of the mirrored disks in a set fails, operations continue using the remaining disk. To re-establish redundancy first break the mirror set.
•
If you no longer want to mirror your drives. This allows you to use the disk space for other purposes.

Note: Although breaking a mirror doesn't delete the data in the set,
you should always back up the data before you perform this procedure.
This ensures that if you have problems, you can recover your data.

In Disk Management, you can break a mirrored set by following these steps:

1.
Right-click one of the volumes in the mirrored set, and then choose Break Mirror.
2.
Confirm that you want to break the mirror by clicking Yes. This creates two independent volumes.

Resynchronizing and Repairing a Mirrored Set

Windows 2000 automatically synchronizes mirrored volumes on dynamic drives.
However, data on mirrored drives can get out of sync.
For example, if one of the drives goes offline, data is only written to the drive that's online.

You can resynchronize and repair mirrored sets on basic and dynamic disks,
but you must rebuild the set using the same disk type. To resynchronize a failed mirror set:

1.
get both drives in the mirrored set online. The status of the mirrored set should read Failed Redundancy. The corrective action you take depends on the status of the failed volume.
2.
If the status is Missing or Offline, make sure that the drive has power and is connected properly. Afterward, start Disk Management, right-click the failed volume and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the volume doesn't return to the Healthy status, right-click the volume and then click Resynchronize Mirror.
3.
If the status is Online (Errors), right-click the failed volume and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the volume doesn't return to the Healthy status, right-click the volume and then click Resynchronize Mirror.
4.
If one of the drives shows as Unreadable, you may need to rescan the drives on the system by selecting Rescan Disks from Disk Management's Action menu. If the drive status doesn't change, you may need to reboot the computer.
5.
If one of the drives still won't come back online, right-click the failed volume and then select Remove Mirror. Next, right-click the remaining volume in the original mirror and then select Add Mirror. You'll now need to mirror the volume on an unallocated area of free space. If you don't have free space, you'll need to create space by deleting other volumes or replacing the failed drive.

Repairing a Mirrored System Volume to Enable Boot

The failure of a mirrored drive may prevent your system from booting.
mirroring the system or boot volume, and the primary mirror diskhas failed.
To correct this problem, you need to replace the failed drive and then use an emergency boot disk^*
for the system or a similarly configured system to enable system boot.
Creating an emergency boot disk is covered in Chapter 14.

Editing BOOT.INI for the Mirror*

Once you have an emergency boot disk, you need to edit the BOOT.INI file
it contains so that the operating system loads from the secondary mirror.
This file contains entries that look like this:

[boot loader] timeout=30 default=multi(0)disk(0)rdisk(0)volume(2)\WIN2000
[operating 
systems]multi(0)disk(0)rdisk(0)volume(2)\WIN2000="Windows 2000Server"

If the secondary mirror drive was on drive 2, you could update the BOOT.INI file shown earlier as follows:

[boot loader] timeout=30 default=multi(0)disk(0)rdisk(1)volume(2)\WIN2000
[operating 
systems]multi(0)disk(0)rdisk(1)volume(2)\WIN2000=""Windows 2000Server"

Note: For a more detailed explanation of BOOT.INI,
see the section of Chapter 10 entitled "Updating the Boot Disk."

Booting and Rebooting the System

Once you update the BOOT.INI file, you can use the emergency boot disk to boot your system.
When the system boots, you'll need to complete the following steps:
1.
Break the mirror set and then re-create the mirror on the drive you replaced, which is usually drive 0. Right-click the remaining volume that was part of the original mirror and then select Add Mirror. Next, follow the technique described in the "Mirroring an Existing Volume" section of this chapter.
2.
When the mirror is completely rebuilt, use Disk Management to break the mirror again. Make sure that the primary drive in the original mirror set has the drive letter that was previously assigned to the complete mirror. If it doesn't, assign the appropriate drive letter.
3.
Right-click the original system volume, and then select Add Mirror. Now re-create the mirror.
4.
Update BOOT.INI so that the original system volume is used during startup.

Removing a Mirrored Set

In Disk Management, you can remove one of the volumes from a mirrored set.
When you do this, all data on the removed mirror is deleted and the space it used is marked as Unallocated.

To remove a mirror, complete the following steps:
1.
In Disk Management, right-click one of the volumes in the mirrored set and then choose Remove Mirror.
2.
Confirm the action when prompted. All data on the removed mirror is deleted.

Repairing a Stripe Set Without Parity

A stripe set without parity doesn't have fault tolerance.
If a drive that is part of a stripe set fails, the entire stripe set is unusable.
Before you try to restore the stripe set, you should repair or replace the failed drive.
Afterward, re-create the stripe set and then recover the data contained on the stripe set from backup.

Regenerating a Stripe Set with Parity

With RAID 5 you can recover the stripe set if a single drive fails.
You'll know that a stripe set with parity drive has failed because the status of the
set changes to Failed Redundancy and the status of the
individual volume changes to Missing, Offline, or Online (Errors).

You can repair RAID 5 on basic and dynamic disks, but you must rebuild the set using the same disk type.
To resolve problems with the RAID 5 set, complete the following steps:
1.
You need to get all drives in the RAID 5 set online. The status of the set should read Failed Redundancy. The corrective action you take depends on the status of the failed volume.
Note: If possible, you should back up the data before you perform this procedure.
This ensures that if you have problems, you can recover your data.
2.
If the status is Missing or Offline, make sure that the drive has power and is connected properly. Afterward, start Disk Management, right-click the failed volume, and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the status of the drive doesn't return to Healthy, right-click the volume and select Regenerate Parity.
3.
If the status is Online (Errors), right-click the failed volume and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the status of the drive doesn't return to Healthy, right-click the volume and select Regenerate Parity.
4.
If one of the drives shows as Unreadable, you may need to rescan the drives on the system by selecting Rescan Disks from Disk Management's Action menu. If the drive status doesn't change, you may need to reboot the computer.
5.
If one of the drives still won't come back online, you need to repair the failed region of the RAID 5 set. Right-click the failed volume and then select Remove Volume. You now need to select an unallocated space on a separate dynamic disk for the RAID 5 set. This space must be at least as large as the region to repair, and it can't be on a drive that's already being used by the RAID 5 set. If you don't have enough space, the Repair Volume command is unavailable, and you'll need to free space by deleting other volumes or replacing the failed drive.

Create an Emergency Recovery Disk

Terse version by DGG adapted from: http://www.microsoft.com/technet/prodtechnol/windows2000serv/maintain/optimize/11w2kadb.mspx

1.	Right-click the simple volume you want to mirror, select Add Mirror.
2.	Use the wizard dialog boxes to select and configure the second volume.

•	If one of the mirrored disks in a set fails, operations continue using the remaining disk. To re-establish redundancy first break the mirror set.
•	If you no longer want to mirror your drives. This allows you to use the disk space for other purposes.

1.	Right-click one of the volumes in the mirrored set, and then choose Break Mirror.
2.	Confirm that you want to break the mirror by clicking Yes. This creates two independent volumes.

1.	get both drives in the mirrored set online. The status of the mirrored set should read Failed Redundancy. The corrective action you take depends on the status of the failed volume.
2.	If the status is Missing or Offline, make sure that the drive has power and is connected properly. Afterward, start Disk Management, right-click the failed volume and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the volume doesn't return to the Healthy status, right-click the volume and then click Resynchronize Mirror.
3.	If the status is Online (Errors), right-click the failed volume and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the volume doesn't return to the Healthy status, right-click the volume and then click Resynchronize Mirror.
4.	If one of the drives shows as Unreadable, you may need to rescan the drives on the system by selecting Rescan Disks from Disk Management's Action menu. If the drive status doesn't change, you may need to reboot the computer.
5.	If one of the drives still won't come back online, right-click the failed volume and then select Remove Mirror. Next, right-click the remaining volume in the original mirror and then select Add Mirror. You'll now need to mirror the volume on an unallocated area of free space. If you don't have free space, you'll need to create space by deleting other volumes or replacing the failed drive.

1.	Break the mirror set and then re-create the mirror on the drive you replaced, which is usually drive 0. Right-click the remaining volume that was part of the original mirror and then select Add Mirror. Next, follow the technique described in the "Mirroring an Existing Volume" section of this chapter.
2.	When the mirror is completely rebuilt, use Disk Management to break the mirror again. Make sure that the primary drive in the original mirror set has the drive letter that was previously assigned to the complete mirror. If it doesn't, assign the appropriate drive letter.
3.	Right-click the original system volume, and then select Add Mirror. Now re-create the mirror.
4.	Update BOOT.INI so that the original system volume is used during startup.

1.	In Disk Management, right-click one of the volumes in the mirrored set and then choose Remove Mirror.
2.	Confirm the action when prompted. All data on the removed mirror is deleted.

1.	You need to get all drives in the RAID 5 set online. The status of the set should read Failed Redundancy. The corrective action you take depends on the status of the failed volume. Note: If possible, you should back up the data before you perform this procedure. This ensures that if you have problems, you can recover your data.
2.	If the status is Missing or Offline, make sure that the drive has power and is connected properly. Afterward, start Disk Management, right-click the failed volume, and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the status of the drive doesn't return to Healthy, right-click the volume and select Regenerate Parity.
3.	If the status is Online (Errors), right-click the failed volume and select Reactivate Disk. The drive's status should change to Regenerating and then to Healthy. If the status of the drive doesn't return to Healthy, right-click the volume and select Regenerate Parity.
4.	If one of the drives shows as Unreadable, you may need to rescan the drives on the system by selecting Rescan Disks from Disk Management's Action menu. If the drive status doesn't change, you may need to reboot the computer.
5.	If one of the drives still won't come back online, you need to repair the failed region of the RAID 5 set. Right-click the failed volume and then select Remove Volume. You now need to select an unallocated space on a separate dynamic disk for the RAID 5 set. This space must be at least as large as the region to repair, and it can't be on a drive that's already being used by the RAID 5 set. If you don't have enough space, the Repair Volume command is unavailable, and you'll need to free space by deleting other volumes or replacing the failed drive.

RAID 0