SAN uses Fibre Channel as the underlying transport mechanism to access block-based devices (disks) using a serialized SCSI protocol over Fibre. SAN can either use arbitrated loops with the FC-AL protocol using hubs to connect the devices or a Fibre Channel switched "fabric" using the FC-SW protocol. Under either protocol, serialized SCSI is still the method of "talking" to the attached devices.
NAS uses Ethernet as the underlying transport mechanism to access file-based devices (a network share) using TCP/IP protocol over either twisted pair copper or optical fiber. The physical medium is irrelevant in either case, but the speed of the transport makes a huge difference in performance. You can access NAS over a simple 10/100 network connection using a standard Ethernet NIC or over Gigabit Ethernet over an optical fiber connection.
NAS is fairly easy to implement, and in most cases it's cheaper since you don't need to purchase all the underlying physical devices for a SAN (switches, cables, HBAs). NAS can be useful for sharing data between operating systems since it looks like a simple file share to the accessing clients. NAS is also a good fit to replace file/print servers.
SAN is better for high performance applications like databases, scientific apps, backup, etc. Anything that needs high performance, low latency access usually runs better on a SAN. NAS and SAN should both be used when architecting an enterprise storage network. Use NAS for file-based access like file/print and Web servers and SAN for the backend stuff. Most SAN vendors let you create a NAS pool from high performing and highly available SAN based storage. Just hook up a NAS head to the SAN and off you go. This will let you redeploy your file/print servers to better use.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum.
This was first published in March 2002