When server disks fill up or the company decides to virtualize, the question of buying central storage comes up, and two terms are heard more than any others: NAS and SAN. The wrong choice means either paying a lot for features you do not need, or buying storage that cannot deliver the speed and availability your key services require.
The Core Difference: File vs. Block
NAS (Network Attached Storage) is a device with its own file system that shares folders over the network using protocols such as SMB (for Windows) and NFS (mostly for Linux and hypervisors). Users or servers work with files, and the NAS itself manages the file system.
SAN (Storage Area Network) presents disk space to servers at the block level, in units called LUNs. The server sees this space as a local disk and formats and manages it itself. The connection is usually made with iSCSI over Ethernet, or with Fibre Channel and dedicated switches and adapters.
In practice, the line between the two has blurred. Many NAS devices also offer iSCSI, and some enterprise storage systems provide both types of access. So instead of looking at the device’s label, look at what kind of access, what level of redundancy and what performance you need.
What Is a NAS Good For?
- Shared folders for users and departments
- Archives of documents, images and large files
- A backup target for software such as Veeam, as long as it is not the only backup copy
- Storing virtual machines over NFS in small environments
The strengths of a NAS are easy setup, web-based management and lower cost. A common weakness of small models is a single controller: if the device’s board or power supply fails, every service that depends on it stops.
What Is a SAN Good For?
- Shared storage for a virtualization cluster, so virtual machines can move between hosts and restart on another host if one fails
- Databases and finance applications with heavy write loads
- Services that need low latency and predictable performance
Enterprise SAN storage usually has dual controllers, redundant power supplies and multiple data paths, so a single component failure does not interrupt service. In return, it costs more, and designing and maintaining it requires more technical expertise.
Take the Storage Network Seriously
Central storage is only as fast and reliable as its network:
- Separate iSCSI or NFS traffic from the user network with dedicated switches, or at least a dedicated VLAN.
- For virtual machines and databases, plan for 10 Gb links or faster.
- Build multiple paths to the storage with two network adapters, two switches and MPIO, so losing a cable or a switch does not cause an outage.
- If you enable Jumbo Frames, the MTU must be the same along the entire path; a mismatch causes strange, hard-to-trace problems.
Common Mistakes
- Putting everything, from user files to virtual machines and backups, on a single-controller NAS
- Keeping backups in a folder that can be deleted with the same domain accounts
- Exposing the NAS management panel to the internet; these devices are a common ransomware target
- Using desktop disks instead of disks designed for NAS or servers
- Not planning for capacity growth and the ability to add disks later
Selection Checklist
- The required access type (file or block) is clear.
- Each service’s tolerance for downtime is known, and controller redundancy is chosen accordingly.
- The storage network is separate and has redundant paths.
- Current capacity and growth over several years have been estimated.
- The management panel is reachable only from the management network.
- An independent backup exists outside this storage system.
ITPlus724 engineers can assess your company’s storage needs and design and deploy the right solution, from a simple NAS to shared storage for a virtualization cluster.