While NAS has pretty much standardized on Ethernet and the Internet Protocol (IP), today's SANs are primarily based on Fibre Channel. But they are also starting to incorporate IP because of cost, interoperability and general ubiquity advantages.

When the round-the-clock global economy took hold, it created a crying need for widely distributed storage-networking architecture. The first version of this distributed architecture was NAS, in which storage subsystems connect directly to a LAN, not a server.

In an NAS implementation, storage can be scaled without affecting other applications within the enterprise. Users can share files across multiple platforms, and it is fairly easy to install and configure a single NAS device.

But while NAS works very well for file I/O transactions, it has performance limitations for block I/O transactions used for most online transaction-processing applications. And manageability suffers as more NAS devices are added.

SANs successfully address these limitations by placing a dedicated storage network in between servers and storage subsystems. In a SAN, any server can connect to any storage subsystem, providing optimum flexibility.

Among SAN protocols, Fibre Channel has captured the highest profile, since it is able to support storage devices as far away as a couple hundred kilometers. Fibre Channel-based SANs can operate in full-duplex mode-sending and receiving data at the same time, while offering centralized management capability, low latency and increased reliability. Still, these SANs are expensive and require specialized skills.

As a result, another approach to building SANs using TCP/IP is gaining traction. Because IP is a mature and ubiquitous network protocol, it delivers a lower cost of ownership and is not plagued by the interoperability problems of Fibre Channel.

There are other advantages as well. For instance, IP-based storage allows the convergence of two independent networks into a homogenous entity. In an IP storage-area network, one Ethernet network is used to access the datacom LAN and storage devices.

In addition, IP storage is easier to manage because it eliminates the need for network administrators to learn new protocols and transport mechanisms.

When interoperability and total cost of ownership take center stage, IP storage is the best bet.

Ron Kroesen is Vice President of Marketing at Startup Silverback Systems (Campbell, Calif.), a Developer of Storage-Network Processors.