Shared Storage Array: What It Is & How It Works

A shared storage array is a centralized pool of disk drives that multiple servers access over a network, appearing to each server as its own local storage. It works by consolidating storage into one managed system and using protocols like Fibre Channel, iSCSI, or NFS to serve data blocks or files to many hosts simultaneously. This article explains the core concepts, components, and practical considerations of shared storage arrays.

Background: Why Shared Storage Exists

Before shared arrays, each server had its own internal disks. This worked for small setups but created problems: storage capacity was wasted (one server runs out while another has spare), failover was difficult (a second server couldn’t easily access the first server’s data), and management was tedious (every disk had to be monitored individually). Shared storage arrays emerged to solve these issues by decoupling storage from individual servers. Instead of buying disks for each machine, you buy one array and connect all servers to it. This enables high-availability clusters, live migration of virtual machines, centralized backup, and simpler capacity planning.

How a Shared Storage Array Works

At its core, an array is a box containing multiple hard drives or SSDs, a controller (or two for redundancy), and network interfaces. The controller organizes the physical disks into logical units using RAID (redundant array of independent disks) or similar techniques, then exposes those logical units to servers. The array presents these units as LUNs (logical unit numbers) in block-based protocols or as shared folders in file-based protocols. When a server writes data, the request travels over the network to the array’s controller, which writes it to the appropriate disk. The controller also handles caching, deduplication, and snapshots, making the array smarter than a simple disk shelf.

Key Components and Protocols

Understanding the parts helps when choosing an array:

Controllers: The brains of the array. They process I/O requests, manage RAID, and provide redundancy. Most enterprise arrays have two controllers active-active or active-passive.
Disks: HDDs (cheap, high capacity) or SSDs (fast, lower capacity). Many arrays support tiering, automatically moving hot data to SSDs and cold data to HDDs.
Network interfaces: Fibre Channel (FC) ports, Ethernet ports for iSCSI, or InfiniBand for high-performance computing.
Protocols: Block-based (FC, iSCSI, SAS) where servers see raw disks, or file-based (NFS, SMB) where servers see shared folders. Some arrays offer both.

For example, a VMware cluster might use iSCSI over 10GbE to connect to the array, while a Windows file server might use SMB 3.0 for shared folders.

Choosing a Shared Storage Array

Selecting the right array depends on your workload and budget. Consider these factors:

Performance: IOPS (input/output operations per second) and throughput. Databases need high IOPS; video streaming needs high throughput.
Capacity: Raw capacity vs. usable capacity after RAID. Account for growth.
Protocol support: Ensure the array supports the protocols your servers use (e.g., iSCSI for Hyper-V, NFS for VMware).
Scalability: Can you add more disks or controllers later? Some arrays scale up (add to one system), others scale out (add more nodes).
Features: Snapshots, replication, thin provisioning, encryption. These add value but also cost.
Redundancy: Dual power supplies, dual controllers, and hot-swappable disks are essential for high availability.

For a small business, a two-bay NAS with iSCSI might suffice. For a large enterprise, a full flash array with synchronous replication across sites might be necessary.

Common Problems and How to Avoid Them

Shared storage arrays can fail, and common issues include:

Performance bottlenecks: Too many servers hammering the array at once. Mitigate by using SSD caching, load balancing, and monitoring I/O.
Network congestion: iSCSI traffic competing with regular LAN traffic. Use dedicated VLANs or separate physical networks.
Firmware bugs: Update firmware regularly, but test in a lab first.
Configuration errors: Wrong RAID level or misconfigured LUN masking can cause data loss. Document everything and test failover.
Single point of failure: If you only have one array, it’s a single point. Consider replication to a second array for disaster recovery.

A common misconception is that RAID eliminates the need for backups. RAID protects against disk failure, but not against accidental deletion, ransomware, or site disasters. Always back up critical data separately.

Costs and Alternatives

Shared storage arrays range from a few hundred dollars for a small NAS to hundreds of thousands for enterprise all-flash systems. Costs include the hardware, software licenses, support contracts, and ongoing power and cooling. Alternatives to a traditional shared array include:

Direct-attached storage (DAS): Cheapest, but not shared. Each server has its own disks.
Software-defined storage (SDS): Uses commodity servers with local disks, pooled together with software like Ceph or VMware vSAN. Scales out easily, but requires more engineering.
Cloud storage: Pay-as-you-go, no hardware to manage. Latency and data sovereignty can be issues.
Hyperconverged infrastructure (HCI): Combines compute and storage in one appliance, managed as a single system. Simpler to deploy than separate servers and arrays.

Each alternative has trade-offs in cost, performance, and management complexity.

Safety and Reliability Considerations

When operating a shared storage array, follow these best practices:

Use RAID 6 or RAID 10 for critical data to tolerate multiple disk failures.
Monitor disk health (SMART) and replace failing disks promptly.
Ensure the array is on a UPS to prevent corruption during power outages.
Test your backups and disaster recovery plan regularly.
Restrict administrative access to the array to authorized personnel only.
Keep firmware and software up to date, but verify compatibility first.

Always consult the vendor’s documentation for specific safety instructions, especially when handling hardware.

Frequently Asked Questions

What is the difference between a SAN and a NAS? A SAN (Storage Area Network) provides block-level access via protocols like Fibre Channel or iSCSI, appearing as raw disks. A NAS (Network Attached Storage) provides file-level access via NFS or SMB, appearing as shared folders. SAN is typically faster and used for databases; NAS is easier to manage for file sharing.

Can I use a shared storage array with just one server? Yes, you can connect a single server to an array, but you lose the main benefit of sharing. It might still be useful for centralized management or future growth.

How do I know how much storage I need? Estimate your current usage and growth rate. Plan for at least 20-30% headroom. Consider thin provisioning to allocate space dynamically, but monitor usage to avoid overcommitment.

What is the difference between RAID 5 and RAID 6? RAID 5 tolerates one disk failure; RAID 6 tolerates two. RAID 6 has more overhead (double parity) but is safer for large disks where rebuild times are long.

Is it safe to use iSCSI over the internet? No, iSCSI is not encrypted by default. Use VPN or IPsec for remote access, or better, use a dedicated private network.

Can I mix different disk sizes in an array? Yes, but the array will treat all disks as the size of the smallest one. For example, mixing 1TB and 2TB disks in RAID 5 will give you capacity based on 1TB disks, wasting the extra space on the 2TB disks.

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.

More cards in this drawer

The whole Numbers and Powers drawer →