While continued growth in subscriptions and increased hunger for bandwidth are good news
for operators on the surface, these factors do not generate commensurate growth in
operator services revenue. Omdia forecasts fixed broadband services revenue in the top 20
global countries will grow from $263bn in 2020 to $299bn in 2025 at a CAGR of 2.6%. This
is a rate far below the expected growth in broadband speeds and far below expected
network traffic growth during that time. While revenue is forecast to grow modestly,
average revenue per user (ARPU) for fixed broadband services is expected to decline at a
1.8% CAGR from 2019 to 2024 based on Omdia’s June 2020 forecast.* At the same time,
global operator capex is expected to rise during this period, increasing at a 2.5% CAGR.†
Constrained revenue translates to constrained ability to spend and imposes two universal
mandates on operators:
- Address continued network growth in the most efficient way possible and at the
lowest possible costs.
- Create profitable new revenue streams to boost growth, either with new services or
by entering new markets.
THE RISE OF OPEN AND DISAGGREGATED NETWORKS
Broadband equipment suppliers have an essential role to play as partners in addressing
these network challenges, but many operators are frustrated by a dynamic that has only
strengthened over the past decade. In a trend that parallels the radio access network
(RAN), the global fixed access network has become highly concentrated among just three
large suppliers. Omdia estimates that 80% share of the $7.8bn global fiber and DSL
broadband equipment market is concentrated among Huawei, ZTE, and Nokia. Concentrated
vendor share limits operators’ freedom to choose and innovate within fixed access.
Operators are increasingly looking to open and disaggregated networking in broadband
access, a trend that has also gained traction in other large networking segments with
concentrated market share. Heavy Reading defines disaggregation in networking as follows:
The separation of networking equipment into functional components and allowing each
component to be individually deployed:
- Ideally, provided in the smallest form-factor capable of delivering a specific
- Equipment should be self-contained, require no additional common
equipment to operate, and incorporate open APIs to enable SDN control.
Equipment should be self-contained, require no additional common
equipment to operate, and incorporate open APIs to enable SDN control.
Benefits of open and disaggregated networks
Heavy Reading sees the following as the primary benefits driving the open and
- Break vendor-proprietary lock-in: Historically, fixed broadband systems have
been closed and proprietary, with PON and DSL hardware, operating system
software, and management software all supplied by the same vendor. Also, as noted,
the number of fixed suppliers has declined as the broadband market has matured,
leaving just three suppliers accounting for 80% of the multibillion-dollar annual
equipment revenue. Disaggregating broadband hardware, software, and
management/control provides the industry an opportunity to break this lock, in part
by lowering the barriers to entry for new entrants. New software suppliers, for
example, can build PON systems by investing R&D exclusively in software
development and using high volume hardware designed and produced by other
- Reduce network costs: The access network accounts for a large share of a Tier 1
operator’s typical network spend, so it is a segment of great attention when it comes
to costs. Opening the fixed access network presents opportunities for new entrants
(as described above), leading to greater competition and thus lower pricing through
market forces. But there are other factors in disaggregated architectures that
contribute to lower costs as well. Disaggregating software components from the
underlying hardware opens new opportunities for running on low cost white box
hardware produced in high volumes. Furthermore, sharing general purpose compute
across multiple functions (i.e., not dedicated solely to FTTx) provides additional cost
savings through greater efficiency.
- Offer new services and monetization opportunities: Broadband ARPUs vary
greatly by country, but the unmistakable trend is that ARPUs (whatever their starting
point) are trending downward. Network operators, universally, see a mandate to
increase revenue. With broadband subscription growth slowing, this growth must
increasingly come from new services and new markets. For fixed broadband, revenue
opportunities span residential, enterprise, and 5G backhaul services. Disaggregation
alone does not result in new services, but open and disaggregated access networks
open the door to high customization and differentiation through in-house
developments as well as third-party contributions.
- Enable faster innovation with diverse ecosystem: Speed of innovation is a
defining characteristic of the hyperscalers, and telecom operators have been
adopting hyperscaler best practices in hopes of emulating their successes.
Disaggregation, open APIs, open ecosystems, and open source development are key
components. One goal is to press the accelerator on telecom innovation, in part to
generate new revenue (as noted above). Significantly, as operators increase their
reliance on nontraditional suppliers, they expect that network innovation will
increasingly be driven by nontraditional suppliers as well.
Status update: Access
Over the past several years, open disaggregation concepts—including SDN and network
functions virtualization (NFV)—have moved to commercial adoption in multiple networking
areas, including data center networks, IP/MPLS, optical infrastructure, the mobile core, and
enterprise software-defined wide-area networks (SD-WANs), among others. Access was not the first focus area for open and disaggregated innovation in fixed or mobile networks due
primarily to the greater size (e.g., number of nodes) and complexity relative to other
segments, such as data center interconnection among a handful of nodes. However,
operators have repeatedly told Heavy Reading—through surveys and one-on-one
interviews—that, while not the first, the access network holds the greatest long-term
potential for disaggregation precisely because of this size and complexity. Capex and opex
savings associated with tens of nodes in a core network is good but applying those savings
to thousands of nodes across an access network is much better.
With early SDN proof points and hands-on experiences gained from other domains,
operators are actively investigating access networks. The trend has advanced significantly
over the past 12 months, with major operators engaging in proofs-of-concept (POCs) and
lab trials. Prominent examples include AT&T, Deutsche Telekom, and Telefónica, among
It is too early to predict the extent to which white box solutions will replace traditional PON
optical line terminal (OLT) solutions. However, several service providers have announced
their interest in more open approaches. In a 2019 white paper making the case for an open
access architecture, Telefónica stated:
This is a necessary evolution in order to deploy 5G, XGS-PON networks and CPEs in a
sustainable way. Our access network will be transformed into an open and standard
based access network. Natively built as a software-based solution with multivendor
components integrated in a whitebox node.*
Technology and market challenges
Disaggregation is relatively new to telecom networks generally, and FTTx is still in a precommercialization stage. In this section, Heavy Reading details the main hurdles that must
be overcome to move to mass-market adoption:
- Developing technology, standards, ecosystem: Technology, standards, and
ecosystems are interrelated, and maturity across all three will be required before
operators migrate to disaggregation en masse. That said, the industry has advanced
in the past 12 months, with ecosystem supplier support growing (including from
application-specific integrated circuit [ASIC] vendors) and operators moving into
- Integrating legacy networks (brownfield vs. greenfield): Not unique to FTTx,
the integration challenge exists for any new technology adoption. An early-stage
solution is to run legacy and new networks in parallel (e.g., coexistence). Umbrella
orchestration of both the legacy and next-gen systems is the most preferable
approach—if legacy systems can support it. Ultimately, operating parallel systems is
difficult to sustain operationally, so operators would need to adopt timelines for
migrating fully to next-gen networks over time. This is, however, never quick.
- Supporting operations and disaggregated networks: The challenge stems from
the fact that disaggregated systems are built from multiple hardware and software
suppliers. A frequent question from operators is: Who is responsible for ensuring operation of the system and stepping in when something fails? Potential solutions are
emerging. One is the hybrid model of the branded white box (or “brite box”) in which
the brand supplier is responsible for the packaged system. Another involves systems
integrator partners, such as Sterlite Technologies Limited (STL), that take on the
responsibility of designing, building, and managing networks for telecom operators.
As the disaggregated technology and ecosystems mature, support options will also
increase in number.
COMPONENTS OF DISAGGREGATED FTTX
This section details the main software and hardware components that make up a
disaggregated FTTx network.
- SDN controller: The SDN controller provides the centralized and automated control
plane function for the data plane elements in the network. It receives and processes
requests from the orchestration layer above through a standardized northbound
interface (NBI), and then communicates those requests to the network elements
below. In a PON network, these elements are OLTs and optical network terminals
(ONTs). Southbound communication is not direct. Rather, an abstraction layer
(described in more detail below) sits between the controller and the elements in
order to simplify the SDN control function.
- Hardware abstraction software: Crucial to the open and disaggregated fixed
broadband model is the insertion of a hardware abstraction layer between the SDN
controller and the data plane elements. The goal of this layer is to hide the
complexity of the data plane from the controller such that, regardless of what data
plane is used, the controller always “sees” and communicates with an Ethernet
switched network. In the SDN-Enabled Broadband Access (SEBA) reference design,
the abstraction functions are labeled “drivers.” In other SDN architectures, software
performing a similar function is often called an adapter. For an FTTx network, drivers
will be needed to support PON, XGS-PON, and NG-PON2. Other types of fixed
broadband networks will need drivers for Gfast, DOCSIS, or other technologies.
- Edge orchestrator: The edge orchestrator is the mediation layer between the
edge/access system and the service provider backend systems (e.g., operations and
business support systems [OSS and BSS] and global orchestration). It communicates
southbound to the SDN controller (or controllers) that manages the access/access
network and northbound to the operator’s backend systems. Multiple operator
OSS/BSS and global orchestration frameworks must be integrated northbound for an
open network. The edge orchestrator function is also called a network edge mediator
(NEM) in Open Networking Foundation (ONF) SEBA terminology.
- Open APIs: APIs are the glue that holds together all the disaggregated FTTx
components, including orchestrators, controllers, and network elements. In the
absence of APIs to allow open communications among the various components, the
communication between them becomes proprietary—which must be avoided in an
- Specifically, the key interface points include the following:
- Edge orchestrator-SDN controller
- SDN controller-hardware abstraction software
- Software abstraction software-forwarding plane/network elements
Many APIs are in play for these functions, including Netconf, gRPC, REST, P4,
OpenFlow, Redfish, and others. Because of their size and influence, hyperscalers
such as Amazon and Google have the power to make APIs de facto standards simply
by adopting them. Individual telecom operators and their suppliers do not have such
influence. Thus, APIs must be endorsed by open source communities (such as ONF
and TIP) and traditional standards bodies (such as the Broadband Forum and
European Telecommunications Standards Institute [ETSI]).
In the case of disaggregated FTTx, there are two main hardware elements, an OLT and an
ONT, that are used in Gigabit Passive Optical Network (GPON), XGS-PON, and NG-PON2
- White box OLT: As in a traditional FTTx architecture, the OLT serves as the service
provider endpoint and resides in a central office (CO) or an edge data center. It
converts electrical signals from the service provider network to the optical signals
transmitted through the PON, and it coordinates the multiplexing of the ONT devices
in the network. White box OLTs are built to open specifications defined in groups
(including OCP) from contract manufacturers such as Edgecore Networks, Celestica,
and CIG. They use merchant media access controller (MAC) and ASIC chips from
chip companies, including Broadcom and Microsemi. Hardware specifications,
including port counts and data rates, will vary depending on form of PON supported.
- White box ONT: As in a traditional FTTx architecture, the ONT is the customer-side
endpoint of a PON network that converts electrical signals from the end user’s
network (residential, business, or wholesale) to optical signals transmitted through
the PON. Just as in the OLT model, a white box ONT is built to an open specification
that can be used by multiple hardware manufacturers. Optical network units (ONUs)
can also be built as pluggable (i.e., “ONT on a stick”) hardware using enhanced small
form-factor pluggable (SFP+) and other pluggable formats.
- Branded white box or brite box hardware: Beyond pure white box
implementation, another deployment model exists in which OLT and ONT hardware is
manufactured as in white box, but a supplier preloads its own software and brands
the devices as its own. The model is a middle ground between traditional hardware
and white box hardware and comes with the benefit of full support from the brite box
Industry ecosystem support
Open Networking Foundation
SEBA is a reference design architecture for a common and modular platform for broadband
access products based on open APIs, white box hardware, and open source software. Within
access, the technology scope is broad and includes PON, XGS-PON, NG-PON2, EPON, future
PON technologies, Gfast, Ethernet, fixed wireless, DOCSIS, and xDSL. Software elements
are built to run in a containerized Kubernetes environment and include a NEM, SDN control,
control applications, access node driver, and aggregation and service gateway driver. For
hardware, SEBA encompasses PON OLTs and ONTs and Gfast distribution point units (DPUs)
in access (with future potential for additional access elements), Layer 2/3 switches and
routers in aggregation, and hardware services to host driver functions.
The SEBA reference design was published in March 2019, and operator champions include
AT&T, Deutsche Telekom, NTT, and Turk Telekom.
A component of the SEBA reference design, Virtual OLT Hardware Abstraction (VOLTHA) is
an open source project to create a hardware abstraction for broadband access equipment.
Today, VOLTHA provides a common GPON control and management system for both white
box and vendor-specific hardware devices. The upcoming introduction of access technology
profiles will extend support to additional access technologies, including EPON, NG-PON2,
and Gfast. On the NBI, VOLTHA works by hiding PON-level details to make the PON network
appear as a programmable Ethernet switch to an SDN controller. On the southbound
interface (SBI), VOLTHA uses different adapters to communicate directly with vendorspecific (i.e., proprietary) and open white box OLTs and ONTs. Additional broadband
hardware types will be incorporated under VOLTHA by creating new southbound adapters.
Operator leads for VOLTHA include AT&T, Deutsche Telekom, Google, NTT, Turk Telekom,
and Telefónica. VOLTHA is expected to reach production by the end of 2020 at AT&T,
Deutsche Telekom, and Turk Telekom.
The Broadband Forum is addressing open and disaggregated networks through several
initiatives under the “Open Broadband” umbrella. Open Broadband-Broadband Access
Abstraction (OB-BBA) is an open source project that specifies NBIs, core components, and
SBI adapters for virtualized broadband access functions. OB-BBA promises to pull together
different access device types, including legacy implementations, under a single network.
The service management and control umbrella will be exposed to management elements,
including SDN management and control and element management systems.
In contrast to the virtualization-focused ONF, OB-BAA is built on the premise that
disaggregation and virtualization are different steps, occurring at different paces according
to operator requirements. For example, operators want to improve the time and resources
required to ensure interoperability between OLTs and ONTs/ONUs from different vendors.
OB-BAA supports this process without requiring virtualization of either OLTs or ONTs/ONUs.
Operator participants include AT&T, BT, CenturyLink, China Mobile, China Telecom, China
Unicom, and TIM. OB-BBA Release 3.0 was released in February 2020.
DISAGGREGATED FTTX USE CASES AND DEPLOYMENT
Disaggregated FTTx use cases are the same as traditional FTTx use cases and include
residential, business/enterprise, and mobile backhaul. Heavy Reading describes each briefly
The rapid and continued increase in download speeds has driven investment in nextgeneration access technologies. As shown in Figure 1 (from the Fixed broadband
drivers/trends section), the global average broadband speed already exceeds 100Mbps
and continues to climb. Copper broadband technologies are quickly becoming obsolete. The
growing trend is to replace the copper access network altogether with fiber-to-the-building/-
Like residential users, businesses also require faster and faster download speeds. One
difference in enterprise services is that businesses are willing to pay more for connectivity,
meaning a faster ROI for operators that offer services. An important consideration in
enterprise applications will be availability and service-level agreements (SLAs), comparable
to SLAs to which enterprises are accustomed in Ethernet and private line services. PONbased enterprise services make sense in converged networks and geographies in which
operators can service both enterprise and residential customers using the same optical
As new macro sites and particularly small cells are deployed globally to support 5G, demand
for backhaul connectivity—moving data from the macro or small cell site to the mobile
core—is set to skyrocket. For small cells in particular, these will be new connections that
require new backhaul infrastructure to support them. As in next-gen broadband access,
fiber is the first choice among operators for 5G backhaul. PON is appealing in converged
networks that share residential and business services but also as an opex-saving option
through the passive optical infrastructure. For 5G, in particular, high data rates are crucial,
but operators are globally are investigating 10Gbps symmetrical XGS-PON as well as fourwavelength NG-PON2. The latter is particularly suited to multiservice converged networks in
which separate wavelengths can be reserved for backhaul services.
At this stage, use cases for disaggregated FTTx map closely to traditional FTTx use cases
(as described above), but disaggregated FTTx differentiates in offering two major
deployment scenarios: a CO scenario and a data center scenario (see Figure 3). Option 1
depicts an XGS-PON or GPON network in which components are disaggregated but
collectively housed in a CO, suitable for CO-based edge point-of-presence (PoP) setup.
Option 2 depicts a disaggregated XGS-PON or GPON in which components are distributed
between the data center and a lightweight “mini CO” that contains OLT functions only,
suitable for regional data center-based edge PoP setup.
CONCLUSIONS AND INDUSTRY NEXT STEPS
Since starting in the data center a decade ago, SDN-based open and disaggregated
networks have been spreading to other areas of networking. It is not a surprise that the
fixed access network has become a key area of operator focus, as access networks hold the
greatest long-term potential to reap the benefits disaggregation. These benefits include the
- Breaking vendor-proprietary lock-in
- Reducing network costs
- Providing new services and network monetization opportunities
- Offering fast innovation cycles with diverse ecosystems
Using open and disaggregated FTTx architectures, operators can meet users’ increasing
bandwidth needs and pursue new opportunities across residential, business, and 5G
backhaul use cases. They can also deploy both traditional and cloud-based architectures as
required. Today, technology, standards, and ecosystems have matured to allow operators to
move from concept to trials to commercial rollouts. A new phase of open and automated
fiber access networking has begun.