Understanding RAID Levels: Choose the Right Storage Solution for Performance and Reliability

1 Disk Groups and Virtual Disks

Modern data centers require ever‑increasing storage capacity. When a single physical disk cannot meet capacity or reliability needs, multiple disks are combined into a disk group, which serves as the foundation for a virtual disk.

A virtual disk is a contiguous storage unit created from a disk group, appearing as an independent drive with greater capacity, higher safety, and data redundancy.

A virtual disk can be:

A complete disk group.

Multiple complete disk groups.

A part of a disk group.

Parts of multiple disk groups combined.

Terminology used later:

Disk group: often called "Drive Group" (DG), "Array", or "RAID group".

Virtual disk: referred to as "Virtual Drive", "Virtual Disk" (VD), "Volume", or "Logical Device" (LD).

1.1 RAID Level Overview

RAID (Redundant Array of Independent Disks) aggregates several physical disks into a single large virtual disk, delivering higher storage performance, I/O throughput, and reliability.

1.1.1 RAID 0

Also known as Striping, RAID 0 offers the highest storage performance by distributing data across multiple disks for parallel access. It provides no data redundancy, making it suitable only for workloads that prioritize speed over data safety.

Processing Flow

When the system issues an I/O request to a three‑disk RAID 0 array, the request is split into three operations, each handled by a separate physical disk, theoretically tripling read/write speed.

1.1.2 RAID 1

Known as Mirroring, RAID 1 writes identical data to two disks simultaneously. It provides high reliability but reduces usable capacity to half of the total.

Processing Flow

Data written to Drive 0 is automatically copied to Drive 1.

Read operations retrieve data from both drives.

1.1.3 RAID 1ADM

Each active disk has two mirror disks. This configuration offers higher reliability than RAID 1 but reduces usable capacity to one‑third of the total.

Processing Flow

Data written to Drive 0 is copied to Drive 1 and Drive 2.

Read operations retrieve data from all three drives.

1.1.4 RAID 5

RAID 5 balances performance, data safety, and cost by using distributed parity. If a single disk fails, the parity information on the remaining disks can reconstruct the lost data.

Processing Flow

Data and its corresponding parity blocks are stored on different disks. When one disk fails, the system uses the remaining data and parity to rebuild the missing information.

Provides better storage efficiency than RAID 1 while offering fault tolerance.

Read speed is slightly lower than RAID 0, but write performance exceeds that of a single disk.

1.1.5 RAID 6

RAID 6 adds a second independent parity block to RAID 5, allowing the array to survive two simultaneous disk failures. This improves reliability at the cost of additional write overhead.

Processing Flow

Data blocks are paired with two separate parity blocks (PA, QA, etc.) and distributed across all disks. The array can rebuild data even if up to two disks fail.

1.1.6 RAID 10

RAID 10 combines mirroring (RAID 1) and striping (RAID 0). It offers the fault tolerance of RAID 1 with performance close to RAID 0.

Processing Flow

Drives 0 & 1 form a mirrored pair (sub‑group 0); Drives 2 & 3 form another mirrored pair (sub‑group 1). Data is striped across the two sub‑groups while each pair mirrors its data.

1.1.7 RAID 10ADM

RAID 10ADM merges RAID 1ADM (triple mirroring) with RAID 0 striping, delivering the reliability of triple mirroring together with high throughput.

Processing Flow

Drives 0‑2 form sub‑group 0 (triple mirror); Drives 3‑5 form sub‑group 1. Data is striped across the two sub‑groups while each sub‑group maintains three copies of the data.

1.1.8 RAID 1E

RAID 1E is an enhanced version of RAID 1 that distributes mirrored data across at least three disks, allowing larger arrays while maintaining redundancy.

Processing Flow

Data is striped across three disks, with each stripe having a mirrored copy on another disk. A single disk failure does not cause data loss.

1.1.9 RAID 50

RAID 50 combines RAID 5 and RAID 0, striping data across multiple RAID 5 groups. It offers both redundancy and high performance, tolerating a single disk failure in each RAID 5 sub‑group.

Processing Flow

Data is striped across several RAID 5 arrays; each RAID 5 array provides parity protection. The overall array benefits from RAID 0’s parallelism.

1.1.10 RAID 60

RAID 60 merges RAID 6 with RAID 0, providing double parity across striped groups. It can survive two simultaneous disk failures within each RAID 6 sub‑group.

Processing Flow

Data and two independent parity blocks are distributed across disks. The array can rebuild data even when two disks in a sub‑group fail.

1.1.11 Fault Tolerance

RAID 0 provides no fault tolerance. RAID 1 offers 100 % redundancy. RAID 5 and RAID 6 use distributed parity to protect against one and two disk failures, respectively. RAID 10, 50, and 60 combine mirroring and parity to achieve higher reliability while preserving capacity.

1.1.12 I/O Performance

RAID 0 delivers the highest I/O speed through striping. RAID 1 incurs a write penalty due to mirroring. RAID 5 and RAID 6 provide good throughput but incur write overhead for parity calculations. RAID 10, 50, and 60 balance performance and redundancy, with performance improving as the number of sub‑groups increases.

1.1.13 Storage Capacity

Capacity varies by level: RAID 0 uses the sum of all disks; RAID 1 reduces usable capacity to the size of the smallest disk; RAID 5 loses the capacity of one disk for parity; RAID 6 loses two disks; RAID 10, 50, and 60 depend on the sum of their sub‑group capacities.