NAS and SAN confusion

The concept of a Storage Area Network (SAN) and Network Attached Storage (NAS) has been around for a long time but it still generates a large amount of confusion amongst technical and semi-technical people alike.

Let’s review:

• SAN

A Storage Area Network (SAN) is composed of an independent system, from one to N) which has the sole purpose of providing access to disk storage via a dedicated network protocol.  That network protocol is typically independent of your normal “computer-to-computer” communications protocol (going over your “Ethernet” network), with one exception (iSCSI).

That’s it, really.  From a computer or server’s perspective, it is storage (not just disk) space made available to it that is not internal to that computer/server.

A SAN does not provide a file system.  This is a common misconception.  It is storage only.  The file system that will be created on that storage is left up to the client.  This is critical to understand when certain advanced features are attempted with a SAN, such as staring storage among several hosts.  Typically, a server, or series of servers, will be put in to place to create, manage, and share the file systems out to clients.

• NAS

Network Attached Storage (NAS) is a way of presenting storage, as a filesystem, to computers and servers.  Most home NAS systems are single units, but, in this discussion, we are talking about the enterprise.  In enterprise solutions, a NAS is typically a localized SAN with a specialized controller in front of it, with the SAN typically bound to the controller/controllers.  The number of devices managed by the NAS is limited, but that limit is extremely high.  The controller is unique in that it does all of what the SAN controller would do plus managing the file systems on that storage.

A NAS does not need another system between it and the hosts that need to access the storage space.  Since all file system management is done at the NAS controller, it will know how to share out the file system to all of its customers.

In Summary: a SAN is not a filesystem

Why Do People Use SANs in the Enterprise?

Cost.  Simply cost.  A SAN solution can be very economical on the cost/gb ratio in the entry-level arena.  The nearly intangible expense to measure is the cost associated to managing the filesystems on the remote servers.  NAS solutions can be quite expensive in the entry-level and lower mid-level arena, and tend to have better ROI in the mid-level and above arena.  However, it all depends on what problem is to be solved.

Secondary answer would be: confusion.  Administrators and managers alike do not fully understand the differences.  Many say “we need more storage space” with the quick answer of “get a SAN” without grasping how the storage will be used and what the trade-offs are (Database? Virtual Machines? etc).  Information which was true in the early days of SANs with regard to the SAN vs. NAS debate no longer holds true, but it still lingers around.

This seems to be a “NAS is better” write-up?

Again, it all depends on the problem that is being solved which should drive the direction of the solution to be implemented. Going by “this one is better than the other” does justice to no one.  The point is, implementing a SAN is not a superior alternative, it is just an alternative.

For example, complex environments which have a need for GlobusFS or MooseFS, would not benefit from a NAS since those highly customized clustered filesystems are not directly supported by a NAS.  Your typical office location made up of windows desktops, servers, and even a mix of linux hosts, would benefit greatly from a NAS.

Remember: A SAN is disk storage for the hosts to manage.  A NAS is storage with a file system.