Who's Eager for 100 Gbps?


This is probably difficult to hear for those of you who are just getting your feet wet with 10 GbE, but that technology is already on the verge of losing its cutting-edge status in favor of even wider band solutions.


Much of the Internet is already looking ahead to 40 GbE within the next few years, with 100 GbE rollouts expected by the middle of the next decade.


New research by the Dell'Oro Group indicates that ISPs are eager to put 40 Gbps in place to accommodate the explosion of video traffic over the Web. That's sure to coincide with a push to install 40 Gbps in the enterprise, both as a means to foster unified fabric architectures that simplify network infrastructures and to ensure that data centers can handle the kind of data growth that most experts predict from the rise of new technologies like virtualization and cloud computing.


The IEEE is already hard at work laying the groundwork for a 40 GbE format for both server and storage applications, as well as a 100 GbE scheme largely aimed at network aggregation. Ratification of both standards is expected by the end of 2010, so we can expect to see actual products within the following year.


Recent demos have already shown that 100 GbE services are feasible through existing network infrastructures. At the recent NXTcomm in Las Vegas, XO Communications, Infinera and Ixia teamed up for a 100 GbE transport system between Vegas and Los Angeles. The system delivered a 100 GbE link from Infinera's DTN switching and transport system through XO's long-haul DWDM network, with monitoring and signal verification by Ixia's XM2 chassis.


If the timeframe for such high-speed networking is too long for you, there are ways to get 100 Gbps throughput and more from current 10 GbE technology. Foundry Networks, which is in the process of being acquired by Brocade, has added network aggregation technology to its NetIron router that lets you combine up to 32 10 GbE trunk ports for an aggregate throughput of 320 Gbps. The company says it uses proprietary load sharing and load balancing algorithms to link network trunks and multiple equal cost multipath (ECMP) routes.


It's probably not fair to describe 10 GbE as an interim solution, considering it does dramatically improve network capabilities above current 1 GbE (or slower) technologies. But it does illustrate the old notion that users will always find ways to push underlying technology to its limits and beyond, leading to a never-ending demand for more, better, faster.


10 GbE can open your network to a whole new level, but it's not the endpoint. Not by a long shot.