Hi there,
Choosing the right enterprise-level data storage technology can be an intricate yet important decision for any organization. This comprehensive guide will simplify things by explaining:
- Key differences between leading networked storage options—NAS and SAN
- How their underlying architectures compare
- Relative advantages for performance, scalability, use cases and budgets
- Help matching infrastructure to your specific storage requirements
Let‘s get started!
Networked Storage 101
Before analyzing NAS and SAN, let‘s briefly understand why businesses need such networked storage in the first place.
As organizations gather more data from customers, operations, IoT sensors etc, centralized and shareable data storage becomes pivotal. Teams need to concurrently access, process and extract insights from burgeoning datasets. Storing terabytes of mission-critical data on individual desktops is risky and restrictive.
Network-attached storage evolved to meet these enterprise data storage demands…
The Rise of Networked Storage
Back in the 90s when networked data storage began to transition from mainframes and proprietary servers to open systems, two primary architectures emerged—file-based Network Attached Storage (NAS) and block-based Storage Area Networks (SAN).
Both enabled consolidating distributed server storage into highly scalable networked data repositories accessible by users and applications via standard protocols. Each approach catered to different types of enterprise workloads leveraging contemporary networking paradigms.
We’ll now examine how these two differ technically.
NAS vs SAN: Key Architecture Differences
NAS and SAN, as abbreviations, sound quite alike. Yet underneath they represent divergent storage infrastructure designs. Let‘s compare some primary technical differences.
| NAS | SAN | |
|---|---|---|
| Topology | Direct-attached network box | Separate FC network just for storage |
| Connected Clients | Any IP network devices | Only FC-enabled servers |
| Access Method | File-level via network sharing | Block-level as direct-attached disks |
| Storage Components | Hardware appliance with disks | Disk arrays + fibre channel directors/switches |
| Truth behind the acronyms | Network-attached storage | Storage area network |
- Topology: Direct vs Separate Network
NAS boxes feature built-in storage connected directly to a standard local area network, just like any other node. This could be an Ethernet LAN or it could even extend over WAN using wide-area file services protocols.
A SAN utilizes dedicated fibre channel networking—high speed switched infrastructure purpose built exclusively for block-based storage connectivity and transport. No other data traffic traverses this network.
- Access Interface: File vs Block
NAS arranges storage in hierarchical file systems, like your pc hard drive. Shared folders called volumes get mounted by clients to access file contents.
SAN disks logically emulate locally attached drives rather than file systems. Servers can directly read/write block storage protocols for low level access.
We‘ll analyze the ramifications of file vs block shortly. But before that, let‘s better understand the key benefits storage networks introduced…
Why Storage Networks?
Direct-attached server storage results in isolated data silos restricting collaboration and causing massive underutilization—some servers are overloaded while other disks lie free. Maintenance, expansion and downtime impact availability. Backup/restore was painful.
Storage networks changed rules of the game for mission-critical data!
Consolidation eliminates stranded storage across scattered servers into mammoth centrally managed pools capable of serving multiple applications simultaneously.
Standardization on leading interconnects like Ethernet, Fibre Channel and Infiniband provides stable data transport independent of ever-changing servers and operating systems.
Virtualization abstracts actual hardware from operating systems freeing applications from local disk restrictions. Features like multi-pathing and snapshots enhance resilience.
Autonomics offloads manual tasks using policy-based optimization and tiering for efficient resource utilization. Diagnostics improve preventative maintenance.
Clearly storage networks address several critical aspects of data lifecycle management…
But while the basic premise is similar between NAS and SAN, the approaches differ greatly. Let‘s analyze this next.
Key Architecture Differences
Earlier we briefly touched upon that core architectural disparity regarding file-level vs block-level access. Let‘s expand on why this matters.
File vs Block
-
NAS File Access
NAS exposes shared directories called volumes mapped over the network to end client devices or servers. Users access data organized into a hierarchical folder constructs using normal file operations like open, read, write and close. Changes are persisted to disks by the NAS head itself.
-
SAN Block Access
A SAN allows servers to directly read and write raw storage blocks exposed by the connected disk arrays or JBODs. There‘s no file system or files per se—the storage appears like locally attached drives without any true ‘network‘ attributes. Changes to block media are managed by the accessing host system.
So why does this file vs block difference even matter?
For starters, files impose structure which adds processing overhead. There‘s the latency of file system metadata lookups plus contention from concurrent client file locks and cache coherency. Throughput suffers from transport protocols ill-suited for efficiency at scale.
Blocks bypass all such constraints via direct memory-like access to disks. Any structure gets imposed at higher levels by server software. The interface transports raw storage streams directly to awaiting applications.
There are certainly tradeoffs to either approach…
Protocol Access
Choosing network interconnect technology induces restrictions on accessibility from various computing platforms. Let‘s examine predominant access protocols:
Common NAS Protocols
-
CIFS – Developed by Microsoft, Common Internet File System enables applications to request files from remote servers on TCP networks. CIFS uses 802.2 Ethernet frames.
-
NFS – Network File System allows UNIX/Linux clients to mount volumes over TCP/IP and UDP networks. Version 4 added security enhancements.
-
AFP – Apple Filing Protocol provides file services for macOS clients. Originally for IP over AppleTalk, AFP now utilizes TCP/IP transport.
Common SAN Protocols
-
FCP – Fibre Channel Protocol enables reliable, lossless transport between servers and storage over dedicated Fibre Channel networks. Provides fabric services like zoning.
-
FCoE – Fibre Channel over Ethernet encapsulates FCP traffic over standard Ethernet allowing use of low-cost switches. Converged network adapters (CNA) connect servers for SAN access.
-
iSCSI – Allows block-based SCSI storage transport using TCP/IP so that SAN can leverage commodity Ethernet infrastructure while preserving FC drive support at the backend.
We can draw some quick protocol inferences already…
- NAS broader client compatibility appeals more for heterogeneous business applications
- SAN requires OS-specific adapter host bus adapters imposed by FC
- iSCSI tries bridging that gap by tunneling SCSI over IP
Beyond protocol gateways, performance and scalability also differ…
Performance and Scalability Compared
Shared storage inherently allows consolidating distributed datasets into mammoth repositories accessible by network connected clients. But support for concurrency and growth varies greatly between architectures.
Scalability
Linear NAS Scaling Challenges
Attaching more NAS appliances indeed adds storage capacity but cascades management overhead. There‘s no unified topology—each box operates discretely, creating network silos. Connecting new devices entails carving out more IP addresses, subnets, adaptors and switch ports. Storage utilization remains fragmented across boxes.
Seamless SAN Scaling
SAN leverages the core advantage of switched networking—scale without hurting latency. More disk arrays stack into the flexible fabric without disruption. Each new port adds to the collective throughput. Device virtualization creates a unified storage pool. Building block modularity simplifies capacity expansion and technology migrations.
Performance
The Latency Factor
We discussed briefly how the file interface adds protocol processing penalties. Beyond that file locks serializing multiple writers, stale data in cached buffers, and contention manifest more severely as usage scales up. Workloads stall unpredictably due to access conflicts.
SAN mitigates such coherency challenges by eliminating files altogether. Direct data transfer via ultra low latency flexible fibre highways keep CPU overhead marginal even under colossal loads. Parallel data access gets governed intelligently at the host level.
Let‘s analyze comparative bandwidth…
Sample Benchmarks
Clearly SAN delivers almost up to 10X throughput under load while cutting latency to nearly half for the same medium. Note that benchmarks use Ethernet for NAS vs Fibre Channel for SAN—rather unfair given the headstart those lines give SAN.
But even if we compare iSCSI to override the networking variable, SAN sustains over 50% higher IOPS thanks to lower protocol overhead relative to file network traffic.
This is the reason SAN remains the undisputed performance choice for large scale-up workloads.
We‘ve covered quite some ground here on the core architectural and technological tradeoffs. Now let‘s shift gears and consider real world usage implications…
Typical Use Cases Compared
We just noted how the innate technological differences make each approach suitable for divergent applications. Let‘s see what that means for practical NAS vs SAN usage scenarios.
NAS Use Cases
Thanks to ease of integration with existing LAN networks, NAS sees ubiquitous deployment for:
-
Endpoint user file shares replacing scattered individual hard drives and USB sticks, introducing data protection and remote access capability limiting disruption.
-
Content repositories serving rich media assets, digital archives and website assets to teams for easier multi-site workflows. Versioning improves collaboration.
-
Low latency analytics where compute remains decentralized but input/output datasets get consolidated on performant scalable NAS like usage telemetry, surveillance data, scientific instruments etc.
-
Cloud gateway for organizations preferring local cache of colder cloud data to mitigate latency and minimize external egress for cost savings given unrestrained NAS capacity scalability.
SAN Use Cases
Given their performance DNA, SAN deployments focus chiefly on:
-
Transactional databases like financial systems, ERP, CRM, records management and other domains dealing with rapid updates and complex relations benefiting from microsecond latency and parallelism.
-
Decision support systems where analytical and reporting queries get served from large historical datasets along expected dimensions leveraging the SAN‘s ability to handle concurrent adhoc access and scale capacity seamlessly.
-
Virtualized servers & desktops that host business applications and workspace images benefit from SAN block storage performance for optimal density and flexible provisioning using technologies like VMware Vvols integrating policy-based automation.
This is by no means an exhaustive list but provides a flavor for typical usage models based on strengths. While NAS targets collaboration, SAN accelerates business processes!
Making a Choice for Your Needs
At this point after going through this wholistic comparison, you likely appreciate their strategic differences. You‘re also better informed to make suitable infrastructure choices matching your objectives.
As a quick rule of thumb selection guide…
| Your Needs | Better Option |
|---|---|
| Simple file sharing and access | NAS |
| Database performance and scalability | SAN |
| Ease of use and setup | NAS |
| Enterprise-class availability | SAN |
| Heterogeneous clients | NAS |
| Virtualized server infrastructure | SAN |
| Video surveillance and streaming | SAN |
| Transaction processing performance | SAN |
| Budget optimization | NAS |
Of course hybrid environments using NAS and SAN together become feasible for larger organizations based on connectivity standards like FCIP and iSCSI. But those build on the fundamental individual strengths covered here.
I hope this guide gave you clarity in simplifying the critical yet complex arena of enterprise networked storage! Feel free to reach out with any other questions.
Summarizing Key NAS vs SAN Differences
Let me leave you with this snapshot view of the major points we covered:
Now over to you. Which data storage network resonates better aligned to your priorities?
Good luck with your deployment!
