SANs VS. NAS: What You Should Know And Why You Should Care.

Though different, they ultimately will merge into Enterprise Storage Networking

SANs and NAS share common objectives: Multiple server storage sharing; simplified storage management; employing network storage to replace direct attached storage; and reducing storage costs. Let's take a look at each to see what they offer and how they compare.

Introducing SAN

A Storage Area Network (SAN) is a new technology designed to overcome throughput and data-sharing issues that are common in existing data networks. SANs are fully scalable, fault-resilient shared data repositories providing unlimited mixing and matching of storage devices, storage space, and even files (under certain conditions) across the enterprise network. SANs are catapulting to popularity as the architecture of choice for centrally managed network storage tasks. SANs are high-speed, high-bandwidth I/O channels (usually Fibre Channel) that connect to the back end of local area network (LAN) servers. They remove cumbersome storage functions off of the server, thus improving overall LAN and WAN performance.

A SAN usually consists of a high-throughput interface (commonly Fibre Channel), interconnects (switches or hubs), and a protocol (IP, SCSI, or IBM's ESCON). A typical SAN includes a blend of storage devices (perhaps automated tape libraries and RAID) capable of communicating with multiple hosts and with each other over a fast, fault-tolerant storage pipeline (the SAN). The SAN is actually a dedicated storage access I/O channel (containing network properties) optimized for handling storage tasks.

Introducing NAS

Network Attached Storage (NAS) is a LAN-attached storage server with a small, optimized system software package dedicated to performing storage-related tasks. It includes pre-configured disk capacity with the ability to be quickly installed on popular network platforms. If you need to easily expand storage capacity while providing uninterrupted high-performance data access across your LAN, you may want to consider a NAS appliance. When budgets are tight and a quick fix is required, a NAS appliance may be a low-cost, efficient solution to dodge that network-to-storage logjam.

View NAS storage as the building block for creating a network with virtually unlimited server storage capacity. NAS accomplishes this expanded capacity without sacrificing either file access speed or system throughput--which would be affected with network storage add-ons such as tape libraries, CD towers, or optical jukeboxes. Instead, NAS storage devices are usually magnetic disks or RAID arrays that provide seamless high-performance file access and filing by operating at full hard disk speed. And the cost in time, resources, and inconvenience is far less than that of increasing internal storage space within the server or implementing a SAN.

Following is an unbiased discussion of the why(s) and when(s) to build a SAN or add-on NAS capability. We close with a peek at how SANs and NAS may be merged to obtain the maximum benefits of these two powerful storage technologies.

The Case For SAN

The need for the SAN architecture came about due to a jump in software and processor performance during the last decade. The business world took advantage of this newfound gain by shifting from data centers to distributed networks, most of which are Unix and NT Windows-based. Though not without improvements of their own, storage devices have not kept up in terms of performance. The difference is felt particularly at sites that typically handle large volume file transfers such as network backup and recovery or constantly deal with high storage access rates, as networks become sluggish and jammed. However, by integrating a SAN infrastructure into an existing network, storage functionality can be scaled to an equal level with software and processor performance.

By extending the any-to-any connectivity of LAN architectures to storage resources, the SAN makes any data on the network, in any location, accessible through multiple paths to any network user, node, or application. These shared-network storage devices are now multiserver accessible and are no longer coupled by restrictive server-to-individual storage unit connections.

The SAN topology frees all storage and file access tasks from the operational limitations of an individual server and the general traffic constraints of a LAN or WAN. Because a SAN externalizes storage outside of a server, the stored data is made available to multiple hosts across multiple platforms without imposing a network overhead burden. Incorporating a SAN into the enterprise network environment enhances LAN/WAN throughput and reduces traffic jams, as the TCP/IP (standard) LAN is freed from large bandwidth-hungry tasks such as data storage and backup, file access, and file retrieval. Immediate performance improvements can be realized in such high throughput, large packet data transfer tasks as data migration, archiving/retrieval, and file read/writes.

The reliability of the storage subsystem is also impacted because of a SAN's fault-tolerant operation. With a SAN, data is accessible through multiple data paths (such as through dual switches or ports). In this way, redundant access can be easily built in. SAN's inherent centralization benefits simplify administration and management of a network's storage components. This centralization significantly improves backup and restore capabilities, making it easier to scale cost-effectively and achieve disaster protection data copies.

The Case For NAS

The NAS architecture was designed with ease of use and heterogeneous connectivity in mind. NAS can be installed almost immediately to handle additional online storage capacity needs. Yet network response time may be an issue. Storage bottlenecks occur when the business encounters a combination of repetitive file accesses, and large program and data files that strain storage resources. With traditional methods of server attached storage, network traffic jams often result because LAN throughput and application performance is limited by the time it takes to read or write across the network. The fundamental issue causing this condition is storage response time (i.e., the time it takes users to gain access to data).

When response to multiple users' requests become "snail-paced," it causes frustration, dissatisfaction with the network (and those associated with maintaining it), and affects the company's ability to serve its customers. NAS devices tackle this problem by providing additional storage capacity at local magnetic hard disk speed across the network. These dedicated NAS devices allow consolidation of multiple traditional file servers into one easy-to-maintain-and-administer unit that is logically attached to multiple subnets or the main network.

NAS provides easily expandable, highly reliable additional storage without requiring network interruption to install. For the MIS professional, this solution comes without the problems usually encountered when increasing server disk capacity--bringing down the server, interrupting users and customers, and risking server reconfiguration problems.

SAN-What You Should Know

Business-critical data is a vital company asset and should be protected as such. Implementing the SAN architecture into an organization's existing computer infrastructure can guarantee the uninterrupted availability and reliability of this important data. The popular Fibre Channel fabric interface often used in building a SAN provides fault-tolerant operations, protecting access to stored data at all times, even upon the failure of a server or other major network component.

Overall network operations are improved by the installation of even a basic SAN. Sites employing bandwidth-hungry applications, such as video, CAD/CAM, or imaging find that the resulting increased bandwidth capabilities greatly enhance overall I/O and network performance. The Fibre-Channel interface provides improved response time on any network running high-end I/O applications such as databases or OLTP. Further improvement in overall network performance can be achieved through zoning and high-availability clusters. Though different from the NAS structure, SAN implementation allows scalable storage that is independent of any specific server or operating system platform on the network.

Basically, the introduction of SAN architecture has brought about more cost-effective, high-availability storage for a wider variety of application environments. Thanks to the electrical isolation of SAN components provided by switches, the hot-pluggable nature of SAN resources allows much easier (and safer) scalability. The SAN dedicated network enables more efficient data management, backup and recovery of lost or corrupted data, and storage node configuration, while preserving ubiquitous data access.

NAS-What You Should Know

NAS storage devices are self-contained with their own embedded software, power supplies, and circuit boards, and so are easy and fast to install. Many can be installed by simply plugging them into an open LAN port without interrupting network service. Unix and Windows NT sites appreciate the almost automatic startup of NAS device insertion. Once initialized, the NAS appears on the network and to applications as another server (the embedded software has detected and configured the existing network), running immediately.

These simple machines have fewer components and software modules because they are optimized to perform a single function--high-performance file access and storage across the network--and are thus extremely reliable. The NAS devices can be monitored through a remote console across the TCP/IP network, a serial console connection, or a GUI using a Java-capable browser. And, an overall boost in system performance can be achieved by off-loading the data access and retrieval chores from the busy general-purpose network servers to the NAS. If the need arises, it is quick and easy to install additional NAS devices. They just "plug-in" as the first one did.

With few exceptions, most user applications find they suddenly have more storage available without ever noticing the addition of the Network Attached Storage file server. The NAS itself provides only file service. Applications continue to run on the local server that accesses the NAS using NDMP (Network Data Management Protocol). NDMP addresses the high overhead and security issues that plagued CIFS and NFS file accesses.

Like the SAN, NAS also enhances the network backup process. Network attached backup processes see the NAS as if it were another local disk. The Unit-type "dump" command can be used to "dump" a whole or partial file system image to tape media or an optical disk device. The NDMP option works in conjunction with Legato's NetWorker or Veritas' NetBackup enterprise-wide backup solution to direct the files on the NAS device to a locally attached backup system controlled tape or similar device.

The Merging Of SAN/NAS Storage Options

As SAN and NAS each have drawbacks that counter their many benefits, creative IT and storage administrators are pulling from both storage techniques in order to implement a storage environment that merges the best of both worlds. The market is pushing for the merging of the SAN and NAS options into something we can call Enterprise Storage Networking.

SANs promise to eliminate the many independent islands of server storage prevalent within most large corporations today and consolidate them into one, centralized heterogeneous storage subsystem. NAS, on the other hand, has the field-proven ability to seamlessly connect storage and share data over a heterogeneous environment. However, its reliability in this area is tempered by its underlying shortcomings in dealing effectively with network bandwidth limitations and overhead as well as online transaction processing (OLTP) loads.

The logical solution to these dilemmas is the merging of the two storage architectures into one storage networking subsystem. This Enterprise Storage Network employs NAS to provide file and data (under specified conditions) sharing over standard network interfaces (e.g., Ethernet or Gigabit Ethernet). It utilizes Fibre Channel SAN to provide dedicated high-speed storage access with greater levels of redundancy to database applications and network storage devices (such as a RAID or an optical library).

This NAS/SAN topology combines the power of NAS (easy, reliable sharing and scaling of storage over the standard network regardless of operating system or server types) and the speed and availability of SAN (access to multiple storage devices over a fast and fault-resilient dedicated Fibre Channel) to formulate a multifaceted data access and handling infrastructure that is capable of meeting all of the enterprise's storage requirements.

By combining SAN and NAS storage architectures, many of the storage application problems facing IT administrators today can be solved. SANs provide for the mixing and matching of storage devices and dynamic reassignment of storage space and units to multiple servers. NAS is well suited for adding capacity to file servers without interrupting network operations. In a merged environment, NAS serves as an "intelligent gateway" to the SAN, enabling end-users speedy and reliable access to their data.

NAS and SAN combine to provide quick storage and capacity scaling, storage consolidation, and LAN-free backup with simplified storage management. By combining the single network-centric administration and file sharing of NAS with the performance and data-centric SAN, storage administrators can meet the most demanding storage application requirements.

Terry Grulke is storage product manager at StorNet, Inc. (Englewood, CO).