Data Redundancy

(OBJ 3.4)

RAID Overview

• RAID (Redundant Array of Independent Disks)
  • Combines multiple physical storage devices into a single logical storage device recognized by the operating system

RAID 0

• Provides data striping across multiple disks

• Used for improved performance but offers no data redundancy

• Multiple drives increase read and write speeds

• Suitable for scenarios where performance is essential, and data redundancy is not a concern

• "The only kind of RAID that does not provide you with any kind of data redundancy"

• Example:

  • Using RAID 0 for a high-end video editing workstation
  • Have two hard disk drives working together so I can work faster

  

RAID 1

• Provides redundancy by mirroring data identically on two storage devices (drives or SSDs)

• Ensures data integrity and availability

• Suitable for critical applications and maintains a complete copy of data on both devices

• Only one storage device can fail without data loss or downtime

• Example:

  • Add two other physical storage devices that are configured as a RAID 1. This way I can use that to store all of my finished videos on.
  • This will ensure I always have an online and on-demand backup ready in case one of my storage devices fails.

  

RAID 5

• Utilizes striping with parity across at least three storage devices

• Offers fault tolerance by distributing data and parity

• Can continue operations if one storage device fails

  • If one disk fails, the other two are going to be able to reconstruct data based on the parity, and that way they can continue their normal operations.

• Data reconstruction is possible but results in slower access speeds

  • The missing portions of data have to be calculated on the fly when the user's requesting the data from the server.

• You can perform a "hot swap" of the failing disk, then the server will rebuild the missing data onto the newly installed drive by calculating all those missing pieces by using the Stripe data pieces and the parity to calculate the missing dataset.

• You can only lose one storage device and still be able to rebuild the data in the RAID, without suffering from any downtime.

  

RAID 6

• Uses data striping across multiple devices with two identical pieces of parity data

  • Striping with double parity

• Similar to RAID 5 but includes double parity data

  • RAID 6 is just one better than RAID 5

• Requires at least four storage devices

• Can withstand the failure of two storage devices without data loss

  • Does give you some more resiliency and redundancy

  

RAID 10

• Combines RAID 1 (mirroring) and RAID 0 (striping)

  • Stripped array of mirrored arrays

• Offers high performance, fault tolerance, and data redundancy

• Requires an even number of storage devices, with a minimum of four

• Can support the loss of up to two storage devices and still be able to rebuild the data without suffering from any downtime, as long as the loss of the storage devices are not both located in the same mirrored array.

  • If you lose one storage device from each mirrored array, that is fine. The RAID 10 will continue to operate without any issues, but you will lost your fault tolerance capability.
  • You want to replace them as soon as possible

  

RAID Resilience Categories

• Failure-resistant
  • Resists hardware malfunctions through redundancy (e.g., RAID 1 and RAID 10)
    • Any RAID using mirroring for data redundancy
  • If one device fails, the system will continue to operate using the data on the remaining devices
• Fault-tolerant
  • Allows continued operation and quick data rebuild in case of failure (e.g., RAID 1, RAID 5, RAID 6, RAID 10)
    • Due to mirroring or striping with parity
• Disaster-tolerant
  • Safeguards against catastrophic events by maintaining data in independent zones (e.g., RAID 1, RAID 10)
    • They have full mirrors of the data on at least one of the storage devices within the array.
  • Disaster-tolerant RAIDs utilize two Independent zones that always have full access to all of the data.

RAIDs summary

• RAIDs are essential for ensuring data redundancy, availability, and performance in enterprise networks
• The choice of RAID type depends on specific requirements for performance and fault tolerance

Configuring a RAID

• Can be easily done in a Windows OS

1. Go to "Disk Management"
2. Initialize unallocated disks

Creating a striped volume

• To stripe across Disk 1 and Disk 2 (RAID 0), right click on Disk 1 and select "New Striped Volume...".
  • Walk through the wizard and add the second disk to it. Then go down "Next"
  • Give the striped volume a drive letter (e.g. D). Hit "Next"
  • Give it a name and click "Next"
  • Now we have Drive D stripping Disk 1 and Disk 2
    • If either one fails, the whole Drive D will fail.

Creating a mirrored volume

• To create a mirrored volume on Disk 3 and Disk 4, right click on Disk 3 and select "New Mirrored Volume..."
  • Walk through the wizard and add the second disk (Disk 4) to it. Then go down "Next"
  • Give the striped volume a drive letter (e.g. E). Hit "Next"
  • Give it a name and click "Next"
  • Now we have Drive E mirroring Disk 3 and Disk 4.
    • Notice you now have this redundant drive, which is a mirrored drive, of type "Dynamic" and only has the capacity size of one of those drives, because one is acting as primary and one is acting as backup.
    • If either of them goes offline, the other will take over.
    • Gives you fault tolerance and creates a 50% Overhead.