The need to provide access to more storage resources characterizes the similarities of both NAS and SAN solutions. Both solutions utilize a network to facilitate this requirement. The NAS and SAN products provide multidevice access that interconnects both servers and storage. With respect to their local operations, they both perform block I/O, as described previously.
The use of a network ties these storage models together and provides a foundation for integration activities. Despite the evolution of data communications networks, the integration of disparate network topologies and protocols remains with us today. Accomplished through the use of bridging mechanisms at the lower levels of the network transport layers , data networks were able to interoperate , allowing the scalability of multiuser networks. The evolution of these elementary bridging technologies to router products facilitated the delivery of client/server data packets as well as sophisticated traffic management and communications reliability.
Although we have yet to see this sophistication within the storage networking world, the use of bridges and routers to accommodate disparate storage protocols in the SAN has been an accepted and widely used mechanism (see Chapter 15). Given their nature of communicating at the file level, NAS can leverage existing bridge, router, and switching technologies as an inherent characteristic necessary to utilize standard Ethernet TCP/IP networks.
In terms of value, they provide similar justification in what they offerthat is, storage consolidation, server consolidation, centralized management, larger storage capacities , multiple application access, and storage scalability. These value propositions center around the effective usage of each storage model, respectively.
This requires the ability to identify and analyze the characteristics of the I/O workload. As discussed in Chapter 17, specific I/O workloads are better suited to either NAS or SAN technologies. Which leads us to the underlying question: Why does anyone need to combine a NAS and SAN configuration? Part of the answer lies in the simple rationale as to why disparate data communications networks need to integrate. The reason is that those users with Macs need to access the same applications and data as PC users. Further, UNIX workstation users operating within an NFS environment need to access shared information on the PC networks. Users of mainframe applications require a separation of complexity from legacy applications that are increasingly integrated with front-end client/server systems. Ultimately, the Internet prompted uniform access to information both locally and worldwide.