RAID (Redundant Array of Independent Disks) is a technology that combines multiple disk drive components into a logical unit. RAID aims to provide increased storage functions and reliability through redundancy. There are several standard architectural RAID levels, each offering specific data availability, performance, capacity and cost benefits.
What is RAID?
RAID is a data storage technology that combines multiple disk drive components into a single logical unit. Data is distributed across the drives in one of several ways called “RAID levels”, depending on the required balance between data availability, performance, capacity and cost.
The different RAID levels provide various tradeoffs between key factors including:
- Data availability – Also known as fault tolerance or resilience. How likely data remains accessible if a drive fails.
- Performance – Read/write speeds, transaction rates and throughput capabilities.
- Capacity – The overall amount of usable disk space in the RAID array.
- Cost – The price of hardware, software and maintenance.
The core goals of RAID are to provide increased data integrity through redundancy while enhancing I/O performance. It aims to minimize the impact of hardware failures and downtime due to drive crashes or related issues. By combining multiple drives into a single logical unit, RAID can recover data if a disk fails by rebuilding the data based on parity information stored across the array.
Core RAID Levels
There are several standard RAID levels, each designed with specific benefits in mind such as performance, capacity or fault tolerance. The most commonly used RAID levels include:
RAID 0 (Striping)
- Minimum Disks: 2
- Data Protection: None
- Performance: High
- Capacity Efficiency: 100%
RAID 0 stripes data across multiple drives without parity information. By striping data across each disk, RAID 0 provides improved performance compared to a single drive. However, it does not provide fault tolerance making it a poor choice for mission critical data. If any drive in the array fails, all data will be lost.
RAID 1 (Mirroring)
- Minimum Disks: 2
- Data Protection: Excellent
- Performance: Medium
- Capacity Efficiency: 50%
RAID 1 mirrors data between two or more disks. When data is written it is duplicated to all disks, providing full data redundancy. RAID 1 provides excellent fault tolerance and easy recovery from drive failure. However, capacity is limited to 50% as data is duplicated on all disks. Performance is medium, with increased read performance but reduced write speeds.
RAID 5 (Block-Level Striping with Distributed Parity)
- Minimum Disks: 3
- Data Protection: Good
- Performance: Medium
- Capacity Efficiency: (N-1)/N
RAID 5 stripes data across multiple disks similar to RAID 0. It also distributes parity information across each drive. This allows reconstruction of data in the event of a single disk failure. RAID 5 requires at least 3 disks. It provides good fault tolerance along with moderate performance and capacity. RAID 5 is commonly used in small to medium storage environments.
RAID 6 (Block-Level Striping with Double Distributed Parity)
- Minimum Disks: 4
- Data Protection: Excellent
- Performance: Medium
- Capacity Efficiency: (N-2)/N
RAID 6 provides double distributed parity, allowing for recovery from the failure of up to two disks. Data and parity is striped across each drive similar to RAID 5. RAID 6 requires a minimum of 4 disks but provides excellent fault tolerance. Performance is medium due to parity calculations. Capacity efficiency is reduced compared to RAID 5.
RAID 10 (Mirroring + Striping)
- Minimum Disks: 4
- Data Protection: Excellent
- Performance: High
- Capacity Efficiency: 50%
RAID 10 combines both mirroring and striping for increased performance and fault tolerance. It provides striping across mirrors, combining the performance of RAID 0 with the redundancy of RAID 1. RAID 10 requires a minimum of 4 disks, provides excellent fault tolerance and high performance. Capacity is limited to 50% due to mirroring.
Nested or Hybrid RAID Levels
In addition to the standard RAID levels, nested or hybrid RAID levels combine two or more RAID levels together for additional flexibility:
RAID 10 (RAID 1+0)
RAID 10 provides a nested RAID 1 inside a RAID 0 array. Disks are mirrored, then striped. This provides the performance benefits of RAID 0 combined with the fault tolerance of RAID 1.
RAID 50 (RAID 5+0)
RAID 50 combines the distributed parity of RAID 5 with the striping of RAID 0. RAID 50 arrays consist of striped RAID 5 sub-arrays, providing increased performance along with distributed parity.
RAID 60 (RAID 6+0)
RAID 60 combines the double parity of RAID 6 with RAID 0 striping. This provides high performance and excellent fault tolerance from the combination of striping and double distributed parity.
Non-Standard RAID Levels
In addition to the standard numbered RAID levels, some vendors have created non-standard proprietary RAID levels. These include:
RAID DP
RAID DP is NetApp’s equivalent to standard RAID 6, providing double parity strips for up to two disk failures.
RAID 7
RAID 7 was an early proprietary RAID level by Storage Computer Corporation that provided optimizations for write-heavy workloads.
RAID S or Parity RAID
A RAID level used by EMC Symmetrix storage systems that provides single parity protection similar to RAID 5.
RAID Z
RAID Z is equivalent to RAID 5 with data protection provided via variable stripe width rather than fixed parity stripes. It was designed for use in ZFS filesystems.
Comparison of RAID Levels
| RAID Level | Minimum Disks | Data Protection | Performance | Capacity Efficiency |
|---|---|---|---|---|
| RAID 0 | 2 | None | High | 100% |
| RAID 1 | 2 | Excellent | Medium | 50% |
| RAID 5 | 3 | Good | Medium | (N-1)/N |
| RAID 6 | 4 | Excellent | Medium | (N-2)/N |
| RAID 10 | 4 | Excellent | High | 50% |
This table compares the different standard RAID levels in terms of minimum drive requirements, data protection, performance, and capacity efficiency. When selecting a RAID level, key factors to consider are the required redundancy, performance needs, and usable capacity.
Benefits of RAID
Implementing RAID provides both performance and data protection benefits for storage environments including:
- Increased data availability – By introducing redundancy, RAID allows continuous access to data if a drive fails.
- Improved I/O performance – Distributing data across multiple disks can provide faster read/write times and transaction rates.
- Minimized hardware downtime – Drive failures can be handled via hot swapping without significant downtime.
- Simplified storage management – Combining drives creates an abstracted logical storage unit.
- Optimized capacity – Capacity can be tailored based on performance and redundancy needs.
For mission critical systems requiring high availability, RAID delivers increased resilience. For high performance applications, it enables improved I/O speeds. The flexibility of RAID levels also allows capacity and performance to be optimized as needed.
Drawbacks of RAID
While RAID delivers important data protection and performance benefits, there are some drawbacks to consider:
- Added hardware cost – Implementing RAID requires additional drives, controllers and in some cases battery backups adding to the cost.
- Complex configurations – Choosing optimal RAID levels and tuning the array requires in-depth technical expertise.
- Capacity overhead – Many RAID levels reduce usable capacity due to parity reserves and mirroring.
- Rebuild times – Rebuilding failed drives can take substantial time depending on the RAID level and size of drives.
- Single point of failure – The RAID controller itself presents a potential single point of failure if not redundant.
While data protection is improved, RAID does not eliminate the need for regular backups and disaster recovery procedures. And additional hardware, complexity and capacity overheads can make RAID more expensive to implement and manage.
Who should use RAID?
Here are some examples of organizations that can strongly benefit from implementing RAID data storage:
- Data centers – Where maximum uptime and availability are critical for hosting customer applications and services.
- Cloud storage providers – To ensure resilience and performance in multi-tenant, highly utilized storage infrastructure.
- Web companies – fast I/O performance allows meeting demands of frequently accessed data and content.
- Financial trading firms – To achieve low-latency transactions and real-time analytics on huge data sets.
- Government agencies – RAID improves storage reliability for critical systems and confidential data.
Essentially any organization where application performance impacts revenue or productivity, or that handles sensitive or regulated data requiring redundancy should evaluate RAID storage.
Getting started with RAID
Some key considerations when getting started with a RAID deployment include:
- Selecting appropriate RAID levels based on capacity, performance and data protection needs.
- Choosing reliable server-grade hard drives designed and tested for RAID environments.
- Utilizing RAID controllers that support required RAID levels, cache memory and connectivity options.
- Understanding how to monitor, manage and tune the RAID array for optimal uptime and performance.
- Ensuring redundancy of RAID controllers to avoid a single point of failure.
- Developing processes to handle drive failures including hot spares and rebuilding.
By designing RAID deployments tailored to an organization’s specific requirements, substantial benefits in availability, performance, and data protection can be achieved. RAID has evolved from humble origins in the late 1980s to a critical data storage technology enabling our modern digital world.
Conclusion
The core RAID levels each provide optimized combinations of performance, capacity, and fault tolerance to meet a variety of data storage needs. Nested RAID configurations further increase flexibility for unique environments and workloads. While introducing increased cost and complexity, RAID delivers powerful data protection and I/O enhancements. For organizations requiring resilient storage infrastructure with minimal disruption, RAID remains an essential technology.