When deploying Windows servers, administrators commonly attach disks directly to each server. For example, the Windows & .NET Magazine environment has separate servers for editorial, production, sales, human resources (HR), and finance, and each of those servers has its own storage.
One problem with this approach is that accurately predicting how much storage each group of users will require is difficult. Consequently, administrators often start out by attaching 100GB of storage, for example, to each server. Within months, however, one department might be using 90 percent of its storage while another is using only 20 percent. Because reallocating storage from one department to another in a Direct Attached Storage (DAS) environment is difficult, most IT departments simply add disk storage as necessary to those servers that need it. According to industrywide research, as much as 70 percent of DAS is typically wasted because it's improperly allocated and too difficult to reallocate.
In addition to the problems of reallocating DAS, the quota-management software that comes with Windows 2000 Server lets you control the amount of storage per user but not the type of files that a user can store. A poll of attendees at a recent Windows & .NET Magazine storage Webcast found that in most cases, more than 30 percent of files saved in DAS environments were non–business-related junk files, such as MP3s. In other words, according to that poll, almost one-third of the disk space that corporate IT purchases and maintains is used for non–business-related files.
Storage Area Networks (SANs) and Networked Attached Storage (NAS) let you easily add or reallocate storage and prevent the storage of nonbusiness file types. SANs are optimized for the high-volume block-oriented data transfers typical of databases and application servers; NAS is optimized for file serving. Consequently, SAN and NAS are often thought of as mutually exclusive technologies. However, SAN and NAS actually complement each other and can be combined to solve common IT storage needs.
SAN vs. NAS
In the simplest terms, NAS is an optimized file server that connects directly to an existing network and serves file data to clients on the network. In contrast, a SAN houses application server data on a dedicated, high-speed network, usually Fibre Channel—based, that serves data to multiple application servers on the dedicated network. Your OS views NAS storage as physical storage, so if you add 500GB of NAS, it appears as 500GB to the OS. SAN storage is divided into logical units, each of which is identified by a LUN. The Windows network sees only the LUNs for the units that you've configured and made available to your network. Table 1 summarizes the major characteristics of NAS and SAN, and Table 2 shows some typical uses for each technology.
But what can you do if—like most midsized to large companies—you have some requirements that would be best satisfied by using NAS and others that seem to call for a SAN? For example, maybe you have old Windows NT servers that handle file serving as well as Microsoft Exchange Server 5.5. You want to migrate both your file and application data, move the appropriate file-access security, create the associated Active Directory (AD) entries, and move the data into Exchange 2000 Server on a Win2K or later server. In this scenario, NAS is the best approach for your file data, and SAN is the optimum solution for your Exchange server application data. In fact, a Windows Powered NAS (WPNAS) device won't let you run Exchange data on it, so your only choices for Exchange data are a SAN or a general-purpose file server.
In such a scenario, you basically have three choices:
- Don't integrate your file and application server data. Keep application server data on a SAN, and keep file server data on multiple separate servers.
- Integrate a SAN and NAS yourself. Install a new general-purpose file server; add Windows Services for UNIX (SFU), Windows Services for NetWare (SFN), and Win2K Services for Macintosh (SFM) as necessary; then add a Fibre Channel host bus adapter (HBA). Most likely, you'd need to provide your own support if you took this route.
- Buy a combined SAN/NAS storage solution. Essentially, this approach combines a SAN with an integrated NAS head. You can buy a combined SAN/NAS solution, or you can get a SAN with an optional NAS head. This option gives you the benefits of both technologies.
Previous
Register
The added complexity of SANs is a significant concern. Typically, only a few people in an IT organization will be able to master the nuances and details of SAN configuration. And although the storage vendors talk a good game about the system being redundant, we have found many instances where the recommendation is to "restart the SAN" after maintenance. If I have many clients/servers depending upon a massive central storage device, they are all affected by any downtime on that device.<P>
Using low-cost NAS technology for disk-to-disk backups is useful. However, it doesn't replace tape for archiving because tape is designed to retain data for decades. Tape cartridges can also be easily taken offsite to provide protection against physical disasters. I understand that special requirements can mandate the choice of a NAS or SAN. However, I think DAS is not getting credit for continuing to be a cost-effective solution for many IT shops.
David Martin January 15, 2004