Fundamentals of Storage Systems: DAS, NAS, SAN, and RAID Technologies

Storage systems are the cornerstone of any IT infrastructure, providing the foundational platform for data persistence and performance.

1. Storage Overview and Methods

1.1 Introduction

Storage refers to the practice of saving data on appropriate media in a safe and efficient manner, ensuring both physical residency and data integrity.

1.2 Three Common Storage Architectures: DAS, NAS, and SAN

1.2.1 DAS (Direct‑Attached Storage)

DAS connects storage devices directly to a server via SCSI or Fibre Channel, similar to typical PC storage. It is suitable for small networks, geographically dispersed sites, and special‑purpose servers, but it consumes host CPU and I/O resources for backup and can become a bottleneck as storage size grows.

1.2.2 NAS (Network‑Attached Storage)

NAS attaches storage to a standard network (e.g., Ethernet) and provides a dedicated file‑storage server. It separates storage from application servers, offering plug‑and‑play deployment, simple management, and lower cost, though it may suffer from lower performance and reliability compared with DAS.

1.2.3 SAN (Storage Area Network)

SAN uses Fibre Channel or IP‑based networks to create a dedicated storage network, enabling high‑speed, scalable, and highly available storage. It consists of storage/backup devices, FC networking components, and management software.

1.3 Comparison of DAS, NAS, and SAN

From a connectivity perspective, DAS offers direct attachment, NAS uses TCP/IP networking, and SAN relies on Fibre Channel. Each has distinct strengths and cost considerations, and all coexist in the market, collectively covering over 70% of disk storage deployments.

2. Disk Arrays and RAID Technology

2.1 Disk Arrays

Disk arrays combine multiple inexpensive disks into a large, redundant storage pool, improving capacity, performance, and fault tolerance through techniques such as parity checks.

2.1.1 Types of Arrays

External disk‑array cabinets (expensive, hot‑swappable).

Internal disk‑array cards (cheaper, require skilled installation, provide online expansion and data protection).

Software‑based virtual arrays (implemented by the OS, lower performance).

2.2 RAID Overview

RAID (Redundant Array of Independent Disks) aggregates multiple disks to achieve higher performance, redundancy, or both. The acronym stands for "Redundant Array of Inexpensive Disks".

Redundant Array of Inexpensive Disks

2.2.1 Common RAID Levels

RAID 0 – Striping only, maximum speed, no redundancy.

RAID 1 – Mirroring, full redundancy, half usable capacity.

RAID 2 – Hamming‑code parity, high redundancy, low speed.

RAID 3 – Byte‑level striping with dedicated parity disk.

RAID 4 – Block‑level striping with dedicated parity.

RAID 5 – Block‑level striping with distributed parity (requires ≥3 disks).

RAID 10 (0+1) – Combination of striping and mirroring, high speed and redundancy.

2.2.2 RAID 0 Details

RAID 0 stripes data across multiple disks, increasing capacity and throughput but offering no fault tolerance. Performance scales with the number of disks, while a single disk failure destroys the array.

2.2.3 RAID 1 Details

RAID 1 mirrors each disk to a partner disk, providing full redundancy at the cost of 50% usable capacity. The array can survive the loss of up to half the disks if each failed disk has a healthy mirror.

2.2.4 RAID 10 Details

RAID 10 combines two mirrored sets (RAID 1) and then stripes them (RAID 0), delivering both high performance and fault tolerance, though it wastes 50% of raw capacity.

2.2.5 RAID 5 Details

RAID 5 distributes both data and parity across all disks, offering good read performance and fault tolerance with a single‑disk failure, but write operations incur additional overhead.

2.3 Hardware RAID vs. Software RAID

Hardware RAID uses dedicated RAID controllers to present a virtual disk to the OS, while software RAID relies on the operating system to manage the array, making the individual member disks visible to the OS.

3. Additional Resources

For further reading, see the original article "Introduction to Storage Systems" and related whitepapers on next‑generation data storage, data‑center networking, and distributed storage architectures.