Hi Ed, a very interesting article, for some not so obvious reasons. While the author's view is accurate to a great extent it is also heavily biased on the precepts of FC. I question the optimal nature of this "paradigm" given the advances now being demonstrated in optical wavelength utilization and higher speed adaptations of Ethernet to same, or other approaches, such as Gigabit Ethernet.
In the case of SANs, GbE is now being used to substitute for FC in some applications a lot more inexpensively due to the embeddedness of its framework since the early Eighties --and the abundance of scalable infrastructure, knowledge and experience that this implies already, in Ethernetworking-- under the direction of emerging forms of bandwidth management, or "traffic cops," some of which are taking their pointers from FC logical underpinnings.
The article obviously enhances one's appreciation of (notice I didn't use the word 'pump') the FC sector, more than it attempts to explore the merits of best fit alternatives. And that's okay, because, at the very least, it's afforded us with a medium for discussion.
It strikes me that the author may be stuck in time, or is heavily influenced by the momentum of an older restrictive technology which strains to move into the future, since he doesn't take the time to mention any other alternatives, save for SCSI which historically has been more limiting than FC from a distance limitation perspective.
I entered into some limited discussion on this matter yesterday on the Gildertech forum with George Gilder, and George makes some very good points about the scalability of these two approaches going into the future.. demonstrating how GbE's higher layer will go to 10gigabits rather easily, leaving FC in the dust at some point. Other considerations have to do with the universality of Ethernet and the narrower scope of FC's features and capabilities, when viewed from a universal networking perspective.
These uses of Ethernet in SANs and NASs are not merely futures to be considered as down the road options. Some of the largest archives I've had the experience in exploring recently now have multiprotocol I-O capabilities (the whole range of higher speed options), left to the selection of the network engineer to pick and choose from, depending on the source of the data's origins.
Something to think about, since 10G is right around the corner, according to many knowledgeable sources I've "spoken" with. Comments and corrections welcome, as always.
Regards, Frank Coluccio |
