In addition, the storage networking architecture is designed to leverage the existing investment organizations have made in networking and storage equipment, personnel, application software, systems and storage management tools, and security. Within eighteen months, the bandwidth of this architecture is expected to increase by a factor of 10 with the advent of 10Gb Ethernet.
DownSouth, I have a few problems with this statement from a technical viewpoint. It needs to be challenged, to some degree, IMHO.
First and foremost, NTAP in particular and NAS in general do nothing to leverage existing investments in storage equipment, application software, and storage management tools. Those three things are built around a 20+y.o. standard protocol called SCSI that defines how "small computer systems interface" to external devices, like disk drives and tape drives. This SCSI protocol runs natively over fibre channel, but not natively over ethernet and IP. Porting existing applications particularly will be an arduous task. It has been attempted before and failed before in plenty of laboratories. Some have been successful, but customers look at it and say "why?" There is nothing to be gained from it. No efficiencies, certainly, and very definitely a degradation in performance.
The 10GbE standard is due in 2002, by the standards guy's own admission. If you are running a going and growing concern, do you want to delay purchases of FAR speedier equipment based on the promises of a standards body? Can you afford to wait for 2002? Many can't and won't...that is why fibre channel, despite all of the typical squabbling by vendors over standards, will succeed in the next two to three years, and I believe longer than that.
To be clear, I have said over and over that the two technologies, NAS and SAN, go hand in hand. Each one complements the other. Ethernet is a networking technology designed for passing huge numbers of small messages back and forth between VERY smart devices. Fibre Channel is an I/O technology designed for high-volume data movement between smart hosts and dumb devices. One day soon, all of this will collapse into InfiniBand, and the distinctions will be essentially meaningless to the average Gorilla investor.
buck |
