SAN, NAS or Both?


In another of our series of technical articles, Feathered Owl gives a few pointers to help you decide whether SAN or NAS is the right storage solution for you.

Over the past twenty years there has been a worldwide migration from host-based to distributed computing. This has had numerous effects, many of them unforeseen when organisations first began augmenting or replacing their mainframe and midrange systems with mini- and microcomputers. One such effect is the increasing move toward storage consolidation and the emergence of Storage as an I.T. discipline in its own right. Today’s I.T. managers have a huge range of vendors and technologies available to them in the Storage arena, and one key decision which must be reached is whether to use Storage Area Network (SAN), Network Attached Storage (NAS), or, indeed, both to satisfy an organisation’s storage consolidation requirements.

Why Consolidate Storage?

Once many organisations had installed dozens, hundreds or even thousands of small computers in their machine rooms, each with their own locally-attached storage, the following issues typically arose: –

  • Storage attached to one computer could not be readily accessed by users or applications on another computer
  • Management of storage resources across the distributed I.T. estate became more onerous
  • Overall utilisation of storage space was inefficient due to the cumulative levels of slack space inaccessible from systems
  • Data backup and recovery became a significant challenge

The Emergence of SAN and NAS

To address these problems, it is now common practice, in I.T. environments of any significant size, to treat data storage as a centralised resource and provide shared access to it via a network. The simplest form of this is the use of file servers to house personal or workgroup data. A file server can be any computer on a network whose storage has been rendered accessible to other computers via a file-sharing protocol such as NFS or CIFS (more on protocols in a while).

File servers became widespread when PC network operating systems such as Netware and Windows for Workgroups emerged; these originally came into being to support sharing of files and printers attached to desktop PCs in office environments. NAS devices evolved from fileservers, as manufacturers introduced dedicated file serving devices to reduce the cost and management overhead associated with multiple server operating systems whose only role in life was to make storage accessible on the network.

At the other end of the spectrum, a more heavyweight solution for consolidating and sharing storage for “enterprise” applications such as databases emerged – the SAN. Historically, a SAN was a means of linking storage devices to multiple computers via a dedicated Fibre Channel network separate from the main data network.

Some More Technical Detail

So a SAN is for big enterprise storage and NAS is for workgroup files and home drives? Yes and no, or maybe not really. We’ll talk about this in a bit. But first there’s some more technical stuff which is important to help us understand the differences between the two technologies. No self-respecting technology is complete without an acronym, so here’s a few explained for you in as simple a fashion as possible: –

SCSI

The Small Computer Systems Interface was ratified by ANSI in 1986 and quickly became an almost universal standard means for attaching storage to mini- and microcomputers via a parallel connection. Other standards which have evolved from this are SCSI-2, SCSI-3, iSCSI and Fibre Channel. As you may be aware, the Fibre Channel network protocol is very important in the world of Storage and SANs.

Fibre Channel

Fibre Channel came about as an alternative to SPI (the SCSI Parallel Interface) which got round the main limitations of parallel SCSI, which are: –

Parallel SCSI cable has length limitations due to crosstalk within copper cables and external interference

  • Parallel SCSI is limited to a maximum of 16 devices on a bus
  • It’s not practical to connect more than one computer to the same storage device

Fibre Channel is a serial protocol which uses fibre optic cable, allowing single cable runs of up 10 kilometres (Fibre Channel can also run over copper cable but that’s another story). As well as supporting longer distances, Fibre Channel supports (in theory) up to 16 million devices on the same bus, meaning that storage devices can be readily shared amongst multiple computers at the network level.

NFS

The Network File System, part of the TCP/IP protocol suite, was developed by Sun Microsystems and released to the public in 1984. Since then it has become the standard means of sharing filesystems in the UNIX world.

CIFS

The Common Internet File System is the commonest protocol used to share files in Windows environments, and is based on Netbios. Functionally, NFS and CIFS are analogous; they just tend to be found in UNIX and Windows environments respectively.

Needless to say it’s all vastly more complex than this, but, in essence, SANs are an evolution of SCSI and NAS is an evolution of NFS and CIFS.

SAN and NAS Defined

Returning now to the issue of distinguishing between SAN and NAS, rather than just considering size and physical architecture, we can use the following definitions: –

A SAN is storage shared at the device level via a serial SCSI protocol such as Fibre Channel

NAS is a computer or device dedicated to sharing files via NFS or CIFS

Technological Convergence in Storage

This definition of SAN and NAS works well when you start to consider that since their inceptions, the uses to which SANs and NAS are put and the hardware on which they are implemented have experienced significant convergence.

As speed, size and reliability have improved, NAS devices have begun to be used for “enterprise” applications such as databases, email and data archiving as opposed to just home drives and shared workgroup directories. SANs have got smaller and more manageable and it is now possible to buy a “SAN in a box” – basically a SAN appliance which contains both storage and SAN fabric to which computers can simply be connected via a Fibre Channel network card or HBA (host bus adapter).

Another recent development is the NAS gateway, which allows SAN storage to be presented via NFS or CIFS over an IP network instead of as a raw chunk or logical unit (LUN) of disk over Fibre Channel.

So what’s the Difference Again?

In summary, the key difference to bear in mind is that SAN is good for making consolidated storage available to multiple computers as raw devices, whilst NAS is good for making it available via network shares. Where an application requires or works best with raw device storage (as is typically the case with enterprise database software), you need a SAN. Where an application will happily access data via a network share, NAS will, more often than not, fit the bill.


One response to “SAN, NAS or Both?”