In a traditional NAS environment, the filer head actually owns that data and that is typically what serves it -- very much like a server-based file-serving environment. If the server or head goes down, you can typically have a passive or a standby node pick it up and serve that same storage.
Traditionally, NAS has suffered from a scalability issue at the higher end and the inability to service multiple concurrent connections. Clustered NAS overcomes these limitations by dynamically distributing client connections to multiple heads. The key thing with clustered NAS is again cost, which will need to be considered in the SMB space.
Check out the entire NAS FAQ guide.
This was first published in July 2008