Sunday, April 23, 2017

The Open Network Automation Platform looks like a turning point for telecom architecture

The biggest news out of the recentOpen Networking Summit in Silicon Valley is that the Open Network Automation Platform (ONAP) Project, which unites AT&T's open source ECOMP and the Open Orchestrator Project (OPEN-O), has made substantive progress toward becoming a dominant lifecycle service management platform for the telecom industry.

Senior executives from AT&T, China Mobile and The Linux Foundation took the ONS stage to declare that ONAP is not only a path towards more open, agile and cost effect carrier networks, but it is perhaps The Path forward for the networks that support the majority of the world’s mobile users. Besides AT&T (with 135 million mobile users) and China Mobile (853 million), the ONAP project has attracted the endorsement, support or commitment to early trials from other key operators including Orange (265 million accesses), Bell, China Telecom, China Unicom and the hot, new Reliance Jio in India. Together, these operators serve 38% of the global wireless subscriber base. The Linux Foundation said they are already in discussions with mobile operators representing a further 34% of the global subscriber base, so clearly we are talking about enormous market clout.

The project description for ONAP states that ONAP is an open source software platform that delivers capabilities for the design, creation, orchestration, monitoring and life cycle management of: low-level virtual network functions (VNFs); the carrier-scale software defined networks (SDNs) that contain them; and higher-level services that combine the above.

From a functionality point of view, ONAP is expected to handle:

•   Automated onboarding.

•   Automated deployment.

•   Automated management.

•   Intelligent operation of network resources using big data and AI for policy optimisation.

The first vendors to announce ONAP support include companies that worked with AT&T on its ECOMP and Domain 2.0 initiatives, including Amdocs, Cisco, Ericsson and Metaswitch, along with key suppliers for the OPEN-O project, which include Huawei, H3C and ZTE). However, with this level of market momentum, expect all vendors to jump on board as quickly as possible. As the ONAP technical steering committees get establish, now would seem a good time to demonstrate enthusiasm and expertise in this new project.

A developer Wiki has already been set-up (and can be accessed here: https://wiki.onap.org/), and the ONAP governing board members have elected the individuals to serve in key roles, while the governance structure was established via The Linux Foundation, with key appointments including:

•   Chair: Chris Rice, SVP of AT&T Labs.

•   President: Yachen Wang, deputy director of the Network Technology Department at China Mobile Research Institute.

•   Treasurer: Vincent Danno, director wireline standards, Innovation Technical and Marketing at Orange corporate.

How we got here

AT&T’s Enhanced Control, Orchestration, Management and Policy (ECOMP) platform has been running in its production network for the past two and half years. The company states that ECOMP contains over 8 million lines of code and that over one hundred virtual network functions are currently supported, including all of its VoLTE traffic. The ECOMP capabilities are also multi-layer, with optical wavelengths orchestrated by the system.

ECOMP was first publicly discussed in depth by AT&T at ONS 2016 (March 2016), when John Donovan, the company’s chief strategy office, described it as the 'next big thing'. Donovan said ECOMP was the most sophisticated software project that AT&T has ever undertaken. ECOMP's aim was to be the engine that powers its software-centric network, with the goal of enabling high utilisation of network resources by combining dynamic, policy-enforced functions for component and workload shaping, placement, execution and administration. At the time, AT&T published its ECOMP white paper, which has since attracted over 7,000 downloads.

In September 2016, Orange became the first telecom company to join AT&T's ECOMP effort. Orange agreed to test the platform for creating and managing its own software-defined network. The carriers also agreed to collaborate on open source and standardisation initiatives to accelerate the standardisation of SDN and NFV. As the first ECOMP partner, Orange perhaps played a critical role in persuading AT&T to move ahead with the idea of contributing its 8 million lines of code to open source. After all, there must have been many reasonable voices from within the company advising against a new and risky strategy of giving away its intellectual property, its crown jewels, for free. In December 2016, Bell Canada became the second carrier to join the ECOMP team.

The OPEN-Orchestrator Project (OPEN-O) was first announced at Mobile World Congress 2016. Its mission was to develop the first open source software framework and orchestrator to enable agile software-defined networking (SDN) and network function virtualisation (NFV) operations. The initiative was driven by China Mobile and Huawei, with early support from Brocade, China Telecom, DynaTrace, Ericsson, F5 Networks, GigaSpaces, Infoblox, Intel, KT, Red Hat, Raisecom, Riverbed and ZTE. In November 2016, OPEN-O launched its first major software release, codenamed SUN.

Many questions about what comes next

With the agreement only just coming into place and technical integration work yet to begin, it is too early to forecast how other models will be impacted. Is it game over for other service orchestration systems? Will other open networking projects, such as ONOS, Open Daylight and OPNFV be redirected to become perfect fits into ONAP? How will ETSI NFV evolve? And what of privately-developed orchestration systems, such as BluePlanet?

Presumably these will also need to adapt to the coming reality of a widely supported ONAP. On the other hand, will one open orchestration system be able to fit all use cases and service providers? And what about the public cloud providers?

The telcos are racing to make their infrastructure hardware more like cloud data centres, while companies such as AWS already have sophisticated software systems in place to bring up new equipment, services and customers quickly and efficiently. From the cloud operators perspective, will the 8 million lines of ONAP prove to be impressive? If so, and if the majority of telcos really agree to this common service orchestration model, will this give them an advantage over cloud operators?


Microsoft has already joined the ONAP project, the only cloud provider to announce its membership in the group so far. However, it is possible that a cloud provider like Google or Microsoft could have a better orchestration system, and perhaps they could find a business case to offer it on a wholesale basis.

See also