AT&T wants to bin 100,000 routers, replace them with white boxes

Carrier tries to speed networking innovation with 'Disaggregated Network Operating System'

AT&T has launched an audacious attempt to push the networking industry towards software-defined networking and white-box hardware.

Revealed in a white paper titled Towards an Open, Disaggregated Network Operating System [PDF], the carrier's plan called for the creation of a “Disaggregated Network Operating System” (dNOS) as part of a push for “A new approach … for router platform development and procurement”. The company also put its money where its mouth is by promising to “evolve its router-platform sourcing process to give preference to dNOS vendors whose products (or committed product-roadmap) are based on using this platform.”

It also revealed that those who buy in to its vision have a big opportunity, by revealing its networks comprise “100,000 interconnected IP/MPLS routers” bought from traditional networking vendors and offering “vertically integrated proprietary hardware and software components.”

The paper's digested argument is that such tight coupling isn't helpful and doesn't make for consistent or easily-evolved networks. The paper's authors have cast their collective eye over technologies including Intel's Data Plane Development Kit, Broadcom silicon and software-defined-networking code, and concluded that if the growing ecosystem of networking hardware and software gets the right help it will mean:

  • Faster introduction of technologies, designs, and features by means of a collaborative ecosystem of hardware and software component vendors
  • Flexibility in network design and service deployment via plug-n-play hardware and software components that can cost-effectively scale up and down
  • Unit-cost reduction through use of standard hardware and software technology components with very large economies-of-scale wherever appropriate

AT&T wants those outcomes very much, but the paper explained it does not believe they can be achieved without a dNOS.

“That ecosystem is only possible if there is a common open platform on which multiple vendors, companies, organizations, and individuals can build on, contribute to, and certify against. Creating an ecosystem of network software and hardware requires a new level of operating system standardization.”

The paper further explained the purpose of AT&T's efforts as follows:

The goal of dNOS is to foster an ecosystem of application and hardware options from multiple vendors. To achieve this vision, it’s critical that both hardware and software include standardized interfaces that a community of developers can coalesce around. A single, standardized NOS is the most efficient and effective means to this end. A single NOS allows for qualification of a common, shared integration infrastructure and APIs to help developers rapidly launch new applications. It allows for ecosystem developers to focus on value adding applications rather than the basic building block components required in all network infrastructure.”

Other key elements of the dNOS include the ability to run on bare metal, or virtualized, and to support both VMs and containers. “At a minimum, it must include support for Intel x86 and ARM,” the paper suggested.

“We call on leading hardware & software vendors to participate in the architecture standardization,” the paper added. “The process will also include, where appropriate, existing industry bodies and standard forums (e.g., Linux Foundation, OCP, OpenConfig, P4, IETF).”

Analysis: 100,000 times wow

AT&T is not alone in feeling that traditional networking vendors aren't particularly helpful any more for those who build networks at significant scales: The Register recently heard Electronic Arts' tech director say such vendors are better at buying lunches than building networks while hyperscalers such as Facebook have written their own software-defined networking tools.

While AT&T's vision allows for multiple vendors to participate in its dNOS project, and for interoperability, it also sends an un-missable signal that proprietary networking hardware's position is perilous.

And that's scary for the likes of Cisco and Juniper, who have spent decades insisting that tightly-coupled proprietary software and hardware are must-haves.

Throw in the carrier's sheer scale, buying power and stated intention to throw its order book around to get the dNOS built, and this white paper will set a snowball rolling.

Where it lands, and who it scoops up along the way, will be mighty interesting to watch. ?


Biting the hand that feeds IT ? 1998–2017