Viodi View – The Bridge Between the Heartland and Silicon Valley

Threat of Disaggregated Network Equipment – Part 3

Introduction:

Disaggregation of IT equipment started with Facebook driving the Open Compute Project (OCP) to open up the design of compute servers. It then extended to bare metal switches and white boxes (such as those from Pica8), which are a huge threat to traditional switch/router equipment vendors.  The threat is now extended to transport/transmission equipment vendors, as described below.

Project CORD for GPON LTE, G.FAST DSL Modems & Mobile Networks:

The latest disaggregated networking effort is to decompose the functional elements of two types of equipment: GPON Line Terminating Equipment (LTE) and for G.FAST (vectored DSL) modems. It’s known as the CORD project, which is an acronym for Central Office Re-architected as a Data Center. Last June, ONOS (a consortium developing an Open Source SDN operating system for service providers) combined with AT&T to demonstrate a CORD Proof of Concept (POC) at the Open Networking Summit which we described in this article.

“One of the ONOS applications that has really taken hold is CORD,” said Bill Snow, vice president of engineering at ON.Lab. “From day one we have targeted ONOS to serve the service provider marketplace… and we found that there was a big hole there,” Snow added.

At Light Reading’s White Boxes for Communications Service Providers event in November, CORD for GPON was described by AT&T’s Ken Duell as “FTTH as a Service.”  It consists of hardware blueprints (schematics?) and open source software modules.

Duell said a CORD-GPON field trial will be held in the 1st Quarter of 2016, building on the PoC referred to above.

In addition to AT&T, ONOS CORD project contributors include Ciena Corp., Ericsson AB, ON.Lab, SK Telecom and Huawei Technologies Co. Ltd.

The New IP recently reported that SK Telecom is working to enhance CORD for the delivery of mobile network use cases. The South Korean telco is leading a project called Simplified Overlay Networking Architecture (SONA), which will ease deployment of software defined data centers, where they have provided OpenStack Switching and OpenStack interfaces for CORD.

Guru Parulkar, executive director of ON.Lab, said the goal of ONOS CORD is to bring the economics of data centers and the agility of cloud to service providers. Guru opined:

“Service providers should be able to build their infrastructure with a few building blocks hopefully built using merchant silicon, white boxes and open source platforms. Telco central offices have to be reinvented because they are where service providers have maximum CAPEX and OPEX spending, but they are also gateways to enable or offer new services to residential and business customers.”

Juniper Disaggregates Junos:

In November, Juniper Networks announced the disaggregation of Junos – it’s network operating system for advanced routing, switching, and security.  This move will allow Juniper’s users to run the software on third-party (bare metal/white box) switches supporting the Open Network Install Environment (ONIE). It also allows customers to install third-party applications such as automation and programming tools or services like deep packet inspection directly on Juniper switches via a VM or container.

Juniper is committed to separating software from hardware as the networking industry shifts to a software focus, Jonathan Davidson, Juniper executive VP and general manager of development and innovation, told attendees at the company’s inaugural NXTWORK customer summit in Santa Clara, California.

Conclusions:

Combine all of the following: price pressures from Chinese network/ telecom equipment vendors (e.g. Huawei and ZTE), carrier consolidation (fewer big gear customers), new competition from the Ericsson-Cisco partnership, and one understands why the telecom equipment market is intensely competitive with razor-thin profit margins. Add “SD-WAN/disaggregation/open source software” to the mix and there is even more of a threat from bare metal switches, white boxes, and commodity transport platforms.

Meanwhile, consortium efforts and industry initiative like CORD/ONOS will surely lead to further minimization of the hardware aspects of large telecom equipment vendors. Software becomes the key factor with most of it going open source (e.g. ONOS and ON.Lab, Open Daylight, Open NFV, ONF, etc).

With that megatrend intact, what role will the big telecom vendors play? How many will be left standing in the next five years is anyone’s guess!

Author Alan Weissberger

View all posts by Alan Weissberger | Website

Posted

in

, , , , ,

by

Alan Weissberger

Tags:

, , , ,

Comments

2 responses to “Threat of Disaggregated Network Equipment – Part 3”

  1. Ken Pyle, Managing Editor Avatar
    Ken Pyle, Managing Editor

    Alan, thanks for the well-thought out and written piece tying together different drivers of both opportunity and consolidation for the remaining telecom equipment vendors. The common denominator that seems to be driving the consolidation is automation. The very tools that the vendors and service providers are providing will continue to mean the disaggregation of their own jobs as well other industry segments. For example, an IoT device likely can report its status more accurately, in real-time and at lower cost, than a human.

    The machines are starting to do the work of the researchers, like those that might have been with Bell Labs in its heyday. Consider the 26 year old who recently developed a self-driving car for $50k ($30k of that was the cost of the car). http://www.bloomberg.com/features/2015-george-hotz-self-driving-car/

    He used Machine Learning techniques to have the car “teach itself” based on his driving patterns. He is going to use another service disrupting technology, Uber, to continue his testing. That is, he plans on being an Uber driver with his car to put up more test miles (that are paid for by his customers – brilliant). He is using $13 smart phone cameras and low-cost mini computers tied into the electronics of a late model automobile.

    The point he makes is that the machine learning software he used eliminated the need for the layers of product definition and translation into code and multiple layers of regression testing. This eliminated what would have taken thousands of people in a traditional product development cycle. More importantly, he had his prototype working in a month and now he (or I should say, the car since it is driving itself) is driving around the streets of San Francisco.

    His prediction is that in 25 years, the last job standing will be that of AI Engineers. He is probably right about the 25 years, but my prediction is that it will be the politicians who manage to be the last ones standing in their jobs. The point is that the consolidation and disaggregation will continue, with fewer and fewer needed to do the work in telecom, as well as most other industries.

    Reply
    1. Alan Weissberger Avatar
      Alan Weissberger

      Ken, thanks for your comment and ominous predictions for telecom/networking jobs.

      OPINION: The current telecom environment of open source software running on commodity hardware leaves little room for companies of any kind to have a “secret sauce” or competitive edge. The only real innovation is occurring at the Application layer— which has nothing to do with telecom or networking. That’s what mobile apps, mobile payments, eCommerce, social networking, and anything as a service is all about—Application layer software rather than the traditional L1-L4 that once characterized telecom and datacom equipment.

      Reply

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.