What is the minimum requirement for RAID 6?

RAID 6 is a type of RAID (Redundant Array of Independent Disks) configuration that uses double distributed parity to provide fault tolerance and protect data against two drive failures. This makes RAID 6 the most reliable and secure RAID level as it can sustain up to two failed drives without losing data. But what is the minimum requirement to set up a RAID 6 array?

Quick Answers

The minimum requirement for RAID 6 is 4 drives. This allows for two drives worth of parity data to be distributed across the array. However, most experts recommend using at least 6 drives for better performance and capacity.

What is RAID 6?

RAID 6, also known as double parity RAID, uses block-level striping with two parity blocks distributed across all member disks. This provides fault tolerance up to two failed drives. The data and parity information are interleaved across the drives at the block level.

In addition to supporting multiple drive failures, RAID 6 offers excellent read performance and write performance comparable to RAID 5. The tradeoff is capacity, as the dual parity requirements reduce the total capacity available for data storage.

How RAID 6 Works

RAID 6 calculates and stores two independent parity blocks known as P and Q. The P parity block works the same as in RAID 5, providing protection against a single drive failure. The Q parity block is an additional calculated block that provides protection against a second drive failure.

If up to two drives fail, the data on the failed drives can be recreated using the P and Q parity blocks and the remaining data blocks. This is known as reconstructing or rebuilding the array. As long as no more than two drives fail at one time, the data remains fully protected and accessible.

Advantages of RAID 6

  • High fault tolerance – Up to two drive failures can be tolerated without data loss
  • Excellent read performance
  • Strong write performance

Disadvantages of RAID 6

  • Higher capacity overhead than RAID 5 due to dual parity
  • Slower write performance than RAID 5 in some workloads
  • Longer rebuild times than RAID 5

Minimum Drives Required for RAID 6

The minimum number of drives required to implement RAID 6 is 4 drives. This allows for the two drives worth of parity data (P and Q blocks) to be distributed across the array.

With only two drives, you cannot have distributed parity, so RAID 6 is impossible. With three drives, you can have a single distributed parity drive for RAID 5, but not the dual parity needed for RAID 6.

Four drives is the absolute minimum that allows the two parity drives to be distributed. This means that in a four drive RAID 6 array, each drive holds some data blocks and some parity blocks.

Four Drive RAID 6 Layout

Here is an example layout of RAID 6 across four drives:

Drive 1 Data 1 Data 2 P Parity 1
Drive 2 Data 3 Q Parity 1 Data 4
Drive 3 Data 5 P Parity 2 Data 6
Drive 4 Q Parity 2 Data 7 Data 8

This demonstrates how the data and dual parity blocks (P and Q) can be distributed across four drives to achieve RAID 6. If any two of the drives fail, the remaining data and parity blocks can be used to reconstruct the data.

Recommended Minimum Drives for RAID 6

While four drives is the absolute minimum required for RAID 6, most experts recommend using at least six drives for a RAID 6 array in practice.

There are several reasons why six drives is a better minimum for real-world implementations:

Capacity

With only four drives, 50% of the total capacity has to be reserved for parity data (two out of four drives). This leaves just two drives worth of capacity for actual data storage. By increasing to six drives, the parity overhead drops to 33% which is much more reasonable.

Performance

RAID 6 performance also improves with more drives. Reads and writes can be distributed across more drives, increasing overall throughput. Rebuild times also improve with more drives to spread the workload.

Redundancy

Having more drives adds redundancy if additional failures occur during a rebuild. With only four drives, if another drive fails during a rebuild there is a high chance of data loss. Six drives can better tolerate an additional failure.

Cost

The cost difference between four drives and six drives is usually minimal compared to the total cost of the storage system. The benefits provided make the small additional cost well worth it.

Six Drive RAID 6 Layout

Here is an example layout of RAID 6 across six drives, demonstrating the improved parity distribution:

Drive 1 Data 1 Data 2 P Parity 1
Drive 2 Data 3 Data 4 Q Parity 1
Drive 3 Data 5 Data 6 P Parity 2
Drive 4 Q Parity 2 Data 7 Data 8
Drive 5 Data 9 P Parity 3 Data 10
Drive 6 Q Parity 3 Data 11 Data 12

This demonstrates the improved distribution of data and parity blocks across six drives. There is less concentration of parity blocks on certain drives. The workload is also distributed across more drives for better performance.

Factors When Choosing Number of Drives

When choosing the number of drives for RAID 6, administrators should consider these factors:

Storage Capacity Needed

The number of drives impacts the total usable capacity after parity overhead. Determine how much storage your applications require.

Performance Requirements

More drives can increase I/O performance. Identify your throughput needs to choose an appropriate drive count.

Availability Requirements

The rebuild window tends to increase as more drives are added. Balance between redundancy and availability.

Budget

Adding more drives increases cost. Balance performance/capacity needs with budget restrictions.

Physical Factor

Consider dimensions of server chassis when planning the number of drives.

Drive Size Considerations

In addition to drive count, the individual drive capacity also impacts RAID 6 performance and efficiency. Some guidelines:

  • More small capacity drives generally provide better performance
  • Fewer large capacity drives reduce rebuild times
  • Match drive capacities to avoid wasted space
  • Balance cost, capacity needs, and rebuild times

RAID 6 Performance Considerations

RAID 6 performance is impacted by the number of drives, drive speed, workload patterns, controller caching, and queued commands. Performance tuning considerations for RAID 6 include:

  • Use faster drive types like SSDs for improved IOPS
  • Add cache to RAID controller to absorb writes
  • Tune queue depths to optimize parallelism
  • Ensure proper I/O balancing across drives
  • Distribute parity evenly across drives

RAID 6 Rebuild Considerations

Rebuild times are an important factor for RAID 6 reliability. During rebuilds, the array is vulnerable to additional drive failures. Guidelines for managing RAID 6 rebuilds:

  • Reduce rebuild times by using fewer, larger drives
  • Monitor drive health to identify failures early
  • Ensure adequate spare drives are available
  • Consider hot spare drives to automatically start rebuilds
  • Pause or throttle rebuilds to reduce impact

Choosing RAID 6 Implementation

RAID 6 can be implemented through:

  • Hardware RAID controller card
  • Software RAID via operating system
  • Firmware RAID inside server
  • Combination of software and hardware RAID

Factors when choosing implementation:

  • Performance – Hardware RAID generally faster than software
  • Cost – Software RAID uses existing server resources
  • Complexity – Hardware RAID easier to manage
  • Flexibility – Software RAID can be reconfigured
  • Caching – Hardware RAID supports caching
  • Compatibility – Software RAID supported across systems

Conclusion

The minimum drive requirement for RAID 6 is four drives, however six or more drives is recommended for most implementations. Drive count, size, speed, workload patterns, and rebuild times should all be considered when architecting a RAID 6 array. RAID 6 provides excellent protection against double drive failures but also requires proper configuration and monitoring to meet availability, performance and capacity needs.