Do we really need 10-GigE?


With one-gigabit/sec NIC cards becoming standard on most computers, it is only a matter of time before we see the rise of the 10 GigE. Some day! A long time ago, I had written a piece on Ethernet as a business model, and argued that with this standard it is all right to waste bandwidth and develop services that cash in on the plentiful yet cheap bandwidth. Ethernet, may not be bespoke, but it is cheap and cheap in the end wins because we live in a WalMart culture. Human nature asks for more and wants to pay less.

Despite that, Nicholas Cravotta, Contributing Technical Editor at Electronics Design News argues that there is a little reason to “supersize” our bandwidth menu. To a large extent I agree, because the true broadband applications are still hard to find. He makes a valid argument that with today’s data – read audio, email, voice, power point presentations, excel spreadsheets etc, the current data speeds are enough to get the job done in a blink of an eye.

Cravotta says the true demand for One GigE is going to come when IP-based video becomes all pervasive, for that is one data set that needs a lot of cheap bandwidth. He argues that since multiple One GigE connections are infinitely cheaper than a 10GigE connection. “10 GE will find a niche as an interswitch connection or link between campuses, although 1 GE seems to be capably serving this need for all but the largest companies,” he concludes.


Papa Tony

I can see an immediate need for even faster connections – Let’s say that Apple comes out with clustering as a simple System Preference in the next operating-system (and it’s very likely – The early version is in public beta right now).

Then, my architectural firm client with four G5’s and three floors of Macs can create a single, 43-processor supercomputer that the guys in design can use to render a new building design, REALLY QUICKLY, even while others on the network are getting their daily work done. Even the secretary’s iMac at the front desk can pitch in.

The biggest bottleneck is the network bandwidth, so gimme more of that!

Chris Garaffa

Papa Tony – You mean something like this: Xgrid? It’s nowhere near complete yet, but it works. Additionally, XCode already has the ability to compile on multiple machines on the same network. We’re almost there.
In general, I think saying we won’t need to go to the next level (in this case, 10GE) because multiple current level (1GE) links are cheaper is only a valid excuse for so long. How many people said the same when the next level was 1GE and the current was 100…

Comments are closed.