Moving from hardware-centric networks to those that rely mostly on software is a monumental task. It is especially difficult for a network as massive as AT&T.
AT&T said this week that on March 28, it completed a trial of a multi-supplier open source white box switch carrying live network traffic. This is a significant step. Vendors to the project were Barefoot Networks, Broadcom, Delta Electronics, Edgecore Networks, Intel and SnapRoute.
Software-defined networks (SDNs) separate the data plane (in other words, the emails, videos and other content) from the control plane (the packets that shepherd the payloads through the network). This segregation creates a lot of possibilities, including network management centralization and operating system standardization. This opens the door to new players, competition and reduced equipment costs.
At the end of the day, it means that networks will run far more efficiently. Imagine a highway teeming with cars. Each has proprietary GPS devices from a small number of vendors. The cars rely on expensive proprietary devices to individually communicate with the satellites that provide directions. In such a world, there is no coordination between cars (except, perhaps between those using the same brand GPS). A high percentage of those drivers may decide to take the same highway at the same time to go to the mall. They would be oblivious if an ambulance needed to use that road.
Now imagine the next-generation network. In the analogy, GPS relies on a common OS. A central traffic controller manages the cars in relation to each other. This is more efficient because the controller would know each cars’ location. Traffic jams would be avoided, shortcuts found for those willing to pay a premium, and the highway cleared for that ambulance. And, since most of the heavy lifting is done by the controller, the GPS device in each car would be simpler and cost less. Finally, since all of these devices now rely on an open source OS available to all, startups would be more common. That would drive costs down further.
Much of this has already been developed in the server world. CIO.com makes the point that the intense tasks switches must perform make the transition more challenging. These devices have to dig deep into packets to determine what priority they have (i.e., whether they are going to the emergency room or the food court), the level of security they need and other attributes. Needless to say, this is a very complex operation that must be done very quickly.
The transition to SDNs poses a threat to the established networking suppliers, as it would to the GPS vendors in the analogy. Until now, a small group of vendors dominated because barriers to entry for new players were daunting. SDNs threaten this neat and tidy landscape, according to Light Reading:
For the makers of proprietary equipment, the ramifications of such a development are scary. Unless they somehow adapt themselves to a white box world, many traditional vendors and service providers could ultimately disappear.
It was reported late last month that Cisco plans to sell its OS on a standalone basis. The Lindt OS will join others that are selling OSes in this manner. This has not been announced. If they do so, however, there are likely two goals: the desire to keep current customers from leaving and to keep influence over the evolution of SDN OSes.
Carl Weinschenk covers telecom for IT Business Edge. He writes about wireless technology, disaster recovery/business continuity, cellular services, the Internet of Things, machine-to-machine communications and other emerging technologies and platforms. He also covers net neutrality and related regulatory issues. Weinschenk has written about the phone companies, cable operators and related companies for decades and is senior editor of Broadband Technology Report. He can be reached at firstname.lastname@example.org and via twitter at @DailyMusicBrk.