RAID 6 is a type of redundant array of independent disks (RAID) that uses block-level striping with two parity blocks distributed across all member disks (Wikipedia). This means data is distributed across multiple drives, with parity information also spread across drives to allow for continued operation even with multiple disk failures.
RAID 6 provides fault tolerance by using dual distributed parity, which means there are two sets of parity data used for data recovery. If a single disk fails, the parity blocks from the remaining disks can rebuild the missing data. If two disks fail, the dual parity blocks can be used to recover data from both failed drives (PCMag).
The key benefits of RAID 6 include increased data protection compared to other RAID levels, the ability to sustain multiple drive failures, and optimized performance through striping data across disks.
How RAID 6 Achieves Redundancy
RAID 6 achieves redundancy through the use of dual parity. This means that RAID 6 uses two separate parity blocks distributed across the array to enable protection against the failure of up to two drives (What is RAID 6? – Definition from SearchStorage). Both of these parity blocks are generated using an XOR operation across the data blocks in the array. If one drive fails, the array can continue operating using one parity block. If a second drive fails, the second parity block provides the redundancy to continue operating and rebuild the failed drives.
This dual parity capability allows RAID 6 arrays to withstand up to two drive failures without experiencing data loss or downtime. This provides a greater level of fault tolerance compared to RAID 5 and most other RAID levels. With the ability to withstand two failed drives, RAID 6 offers excellent protection for large storage arrays where the probability of multiple drive failures is higher (RAID 5 vs. RAID 6: Capacity, performance, durability). The dual parity comes at the cost of write performance and overall storage capacity versus RAID 5, but provides significantly improved redundancy and reliability.
Drive Capacity
RAID 6 requires a minimum of 4 drives to be implemented. The maximum number of drives supported depends on the RAID controller, but is typically between 16-32 drives. According to Microsemi, “RAID 6 requires a minimum of 4 drives and a maximum of 32 drives to be implemented.”
When determining the maximum drives for a RAID 6 array, Spiceworks recommends considering the size of the individual drives rather than just the number of drives. They note that they have “run RAID5 on up to 900GB drives with little to no issue.” So while 32 drives may be the technical maximum, the practical limit may be lower depending on the storage capacity of each drive.
Performance
RAID 6 offers good read performance since data can be read in parallel from multiple drives. However, write performance suffers compared to RAID 5 and RAID 10 due to the double parity calculation. Each write requires the parity data to be updated on two drives, which adds significant overhead.[1]
Specifically, RAID 6 write performance is about half of RAID 10 and can be up to 25% slower than RAID 5. One study found that RAID 6 performance got progressively worse as more disks were added, making large arrays significantly slower for writes.[2]
Rebuild times are also slower with RAID 6 compared to RAID 5. If a drive fails, the array must recalculate parity across all remaining drives, which takes longer with the dual parity of RAID 6. However, rebuild times are still faster than restoring from backups.[3]
Cost Considerations
RAID 6 incurs additional costs compared to single disks or other RAID levels due to requiring a minimum of 4 drives. At least 2 drives are needed for redundancy, with each additional drive adding more storage capacity but also more expense. According to one analysis, the overhead cost of RAID 6 can be 20-33% compared to a single disk [1]. This means you get 67-80% of the raw storage capacity of the disks.
The controller is another significant cost factor for RAID 6. You need an advanced RAID controller capable of the dual parity calculations required. Controllers with RAID 6 capability tend to cost over $200. The controller cache memory also impacts performance and potentially cost. In addition, some server-grade motherboards have RAID controllers built-in which can help reduce hardware expenses.
Data Protection
RAID 6 provides robust data protection through the use of dual parity. This allows RAID 6 to continue operating normally even if two drives fail simultaneously 1. The dual parity allows the array to recalculate the data that was on the failed drives and redistribute it across the remaining good drives.
However, if more than two drives fail at the same time before the failed drives are replaced, irrevocable data loss could occur. Proper monitoring and timely replacement of failed drives is critical to avoid this scenario. RAID 6 provides excellent protection against most failure scenarios, but does not completely eliminate the risk of mass drive failure leading to data loss 2.
Compatibility
RAID 6 is compatible with most modern operating systems, including Windows, Linux, and macOS. The main requirements for RAID 6 compatibility are:
Hardware RAID Controller – Most RAID 6 implementations require a dedicated hardware RAID controller that supports the RAID 6 level. Many RAID controllers from companies like LSI, Adaptec, and HighPoint offer RAID 6 support.
Firmware Support – The system firmware and RAID controller firmware must have driver support for RAID 6 configurations. Most current firmware versions support RAID 6.
Operating System Drivers – The OS needs RAID drivers installed that are compatible with the RAID controller being used. Most major OSes like Windows, Linux, and macOS include built-in RAID 6 drivers or compatible third party drivers.
Disk Interface – The physical hard disks must connect to the RAID controller using a compatible interface like SATA, SAS, or NVMe. Mixing disk interfaces in one RAID 6 array is not recommended.
Disk Capacity – There are no specific capacity limits for RAID 6 compatibility, but very large >8TB drives may have compatibility issues on older RAID controllers.
The Reddit and IBM links point out that certain older hardware, especially low-end SATA RAID controllers, may lack full RAID 6 compatibility. Overall though, RAID 6 support is widespread on most modern hardware.
Alternatives to RAID 6
While RAID 6 offers excellent redundancy and protection against multiple drive failures, there are some alternatives that may better suit certain use cases:
RAID 10
RAID 10, also known as RAID 1+0, combines mirroring and striping for redundancy and performance. Data is mirrored across pairs of drives and then striped across multiple sets of mirrored drives. RAID 10 can sustain multiple drive losses as long as no more than 1 drive fails per mirrored set. It offers better performance than RAID 6 but at a higher cost. RAID 10 is commonly used for transactional databases and other I/O intensive applications where performance is critical (Source).
RAID 5
RAID 5 uses distributed parity and striping to provide redundancy while using less capacity than RAID 6. It can sustain a single drive failure but is at risk if a second drive fails before the first failed drive is replaced. RAID 5 has largely been superseded by RAID 6 due to the higher likelihood of UREs on larger drives. However, RAID 5 may still be suitable for use cases with smaller drives or less critical data.
Non-RAID Options
For use cases that do not require redundancy, non-RAID options like JBOD or individual drives may be suitable. These options maximize storage capacity and performance but do not provide any protection against drive failures. Regular backups are essential when using non-redundant storage configurations.
Use Cases
RAID 6 is well-suited for certain storage use cases where redundancy and protection against disk failure is critical. Some key examples include:
Media storage – Media production environments like video editing or animation studios require secure and redundant storage for large video and media files. RAID 6 provides protection in case multiple drives fail, preventing loss of irreplaceable media assets.
Financial data – Banks, investment firms, and other financial organizations often use RAID 6 for transactional databases, trading platforms, and other mission-critical financial data where loss is not an option.
Other suitable applications – RAID 6 can also be advantageous for medical/healthcare records, government/military data, virtualization, email storage, and other use cases where high availability and data protection are essential.
The redundant nature of RAID 6 makes it well-suited for large storage arrays with 10 or more drives. The more drives, the higher likelihood of multiple disk failures over time, so RAID 6 provides excellent protection.
Conclusion
In summary, RAID 6 is a redundant array of independent disks (RAID) configuration that uses dual parity to protect against two disk failures. Some key points about RAID 6 include:
- RAID 6 can support a minimum of 3 drives and a maximum of 128 drives.
- It achieves fault tolerance by writing data and parity information across all the drives.
- Performance is slower than RAID 0 or RAID 5 but faster than a single drive.
- RAID 6 provides excellent protection against data loss in the event of multiple drive failures.
- The cost is higher than other RAID types since more drives are needed for redundancy.
Overall, RAID 6 is recommended for mission-critical storage that cannot afford data loss. The dual parity provides excellent protection, though at a higher cost. For less critical data storage needs, RAID 5 or RAID 10 may provide sufficient redundancy at a lower cost.