Storage Technology News:
Tutorial on network-attached storage
02 Jan 2008 | SearchStorage.com
Network-attached storage is a class of dedicated hard disk-based file storage devices. Rather than storing data on disks allocated on file servers scattered throughout an organization, a NAS device provides central, consolidated disk storage available to LAN users through a standard network connection.
Most NAS systems receive an IP address, connect to the LAN through an Ethernet cable, and reside on the LAN as an independent network device. NAS devices can also be part of a SAN. Although IP is the most common network protocol, many NAS products support other network protocols, such as Novell Inc.'s IPX and Microsoft's NetBEUI. NAS products can share and exchange files using established file sharing protocols such as Sun Microsystems Inc.'s NFS or Microsoft's CIFS open standard. Other protocols allow NAS "boxes" to operate across a wider variety of network infrastructures.
NAS devices offer storage administrators the abilities to expand and consolidate their storage infrastructures. Network storage no longer relies on disks in a local server. And since storage is no longer limited by the number of disks that a server can hold, a NAS product can hold numerous disks (and can easily include enough disks to support RAID), and multiple NAS "boxes" can be attached to the network for extensive storage expansion. In most cases, a single NAS box can replace numerous file servers, resulting in significant consolidation. If one NAS box runs short on storage, more NAS boxes can easily be installed.
For instance, the FAS270 from Network Appliance Inc. offers up to 16TB of raw capacity using up to 56 disk drives in a 3U 19-inch rack mount. With user storage taken out of the local server, network users don't demand the server's processing time for mundane storage tasks, which in turn improves performance of local application servers. NAS systems also include some onboard RAM to cache network data to or from the disks. Small NAS devices may only provide a 128 MB to 256 MB cache; enterprise-class NAS systems may offer cache of 8 GB or more.
Although some NAS boxes will run a standard operating system like Windows, many NAS devices run their own proprietary operating system. For example, the NAS platforms from NetApp use the company's proprietary Data ONTAP operating system. NAS platforms are typically managed and configured using integrated software utilities that run across any standard Web browser. This allows storage administrators to check NAS status, diagnose performance issues and make changes to the NAS configuration from any workstation on the LAN. Management utilities are able to manage heterogeneous multiple NAS boxes as more storage is added to the infrastructure, easing the management burden on storage administrators.
NAS systems are denoted by their drive support, the total number of drives and total capacity/expandability. SCSI drives are one common drive type and SAS drive compatibility is available, but by far the most popular drive type supported at this time is SATA, which offers low-cost bulk storage. Workgroup-type NAS systems generally support 1 TB and up of disk storage. Basic NAS units employ a group of four to six hard disks, though some models support expansion disk racks that allow extended storage up to 30 TB. Enterprise-class NAS systems can implement many disks, to achieve capacities well over 100 TB.
The network interface can become a LAN bottleneck for storage traffic. Since a NAS can replace many individual file servers, the storage traffic that had previously been distributed throughout the data center or organization is now centralized in one system. In other words, the connectivity must scale to service the concentration of storage. Many NAS products support Gigabit Ethernet (GigE) for faster data transfers across the network. Some NAS products provide dual (or even quad) GigE connections for improved network performance and interface redundancy (e.g., failover or port aggregation).
As NAS technology extends further into the enterprise, the need for performance also grows -- establishing the notion of "high-performance NAS." Substantial NAS deployments need more throughput to pass ever-growing volumes of data between disks and the network. This means better internal disk controllers for more IOPs and additional connectivity to support higher network traffic. Most high-performance NAS platforms employ global file systems for extensive data access, and the systems are often tuned to optimize data streams for specific applications such as transactional data handling or media file streaming. NAS systems can sometimes achieve better performance by clustering NAS nodes together, but clustering alone is no guarantee of added performance.