Facebook Telecom Infrastructure Project starts chucking rocks at mobile model
Targets the heart of the network
Analysis Nine months after announcing its Telecom Infrastructure Project (TIP), Facebook has held its first summit and unveiled new partners and a first concrete project, a white box transponder/router for fibre backhaul, called Voyager.
This is an indication, if any were needed, that the social media giant was not just tub-thumping when it pledged, at this year’s Mobile World Congress, to shake up the traditional network infrastructure supply chain.
TIP membership expands, though no US giants
The first TIP Summit was held earlier this month in Menlo Park, California. New operator members include Bell Canada, Middle Eastern carrier du, NBN and Telstra from Australia, Orange and the Nordics’ Telia. New vendor sign-ups came from Accenture, Amdocs, Canonical, HPE and Toyota InfoTechnology Center, among others.
Founder members included Nokia, SK Telecom, Deutsche Telekom, Intel and others, and in May they were joined by more major operators - Axiata Digital, Indosat, MTN, Telefonica and Vodafone – plus additional vendors.
There are still significant gaps, notably the two US giants Verizon and AT&T (though they are part of Facebook’s similar, older initiative for data centre gear, Open Compute Project). TIP is a double-edged sword for those operators which have been accustomed to tight control of their circle of suppliers, since in return for the promise of lower costs, they hand the initiative in driving network infrastructure to Facebook – a company which, on the services side, is eating their lunch with over-the-top services such as Messenger and WhatsApp.
But TIP now has about 300 members and has enough representation from major carriers to be capable of significant disruption to business as normal for the vendors, on a near-global basis.
As with Google Android on the open software side, most of the technologies in TIP are, so far at least, coming from Facebook itself. That gives it a large level of control of how platforms evolve, but also leaves it with the bulk of the cost. Its latest contribution is Voyager, which it says is the first white box transponder/router, and which supports Open Packet DWDM (dense wave division multiplexing*). This design will be available to the TIP operators and vendors via one of its sub-projects, called Backhaul: Open Optical Packet Transport.
Facebook debuts Open Packet Transport
Facebook is already trialling its implementation of the Open DWDM platform, Voyager, which is designed to drive down costs in the same way as Facebook’s Wedge 100 white box switch did in the data center. Voyager uses the same Broadcom Tomahawk switch ASIC chip as Wedge 100. Facebook said that other partners in developing Voyager were Acacia Communications, Lumentum Holdings, Celestica and Snaproute.
The first triallists are Equinix and MTN in South Africa. Equinix said preliminary trial results "showed zero packet loss and significant overall cost savings due to this disaggregated hardware and software networking model”.
The first vendor to support the Open Packet Transport platform is ADVA Optical Networking, which plans to sell Voyager and pro-vide network management, operations support and maintenance. ADVA CTO and COO Christoph Glingener called Voyager: “A game changer that will open up networks to a whole new range of customers”.
All this is not just about cost, but about keeping up the rate of evolution in important technologies such as DWDM*, which enabled a step increase in fiber network capacity when it was first commercialized. In a blog post, three Facebook engineers, Ilya Lyubomirsky, Brian Taylor and Hans-Juergen Schmidtke, wrote: “The pace of innovation has slowed over the past 10 years as we approach the limits of spectral efficiency”.
They argue in the post that an open approach drives greater efficiencies into DWDM transport. "By unbundling the hardware and software in existing 'black box' systems, which include transponders, filters, line systems, and control and management software, we can advance each component independently and deliver even more bandwidth with greater cost efficiency," they wrote.
Voyager is similar to many other small form-factor WDM units available in the competitive metro area fibre market, so will have to live up to its promises of low costs to make an impact, beyond Facebook’s own deployments, and to drive adoption of the open platform. It is an open one-rack unit DWDM device with eight 100Gbps quad small form-factor clients and four 200Gbps 16QAM on the line side. It can support long-haul transport and metro area data centre links.
It may not be a revolutionary project per se, but the optical space is ripe for change in backhaul and data centres. The optical transport network is one of the areas where telcos are moving most rapidly towards software-defined networking (SDN), which is a major spur to migrate from proprietary equipment to off-the-shelf boxes.
Google and Facebook blow apart network norms
Along with Google, Amazon and Microsoft, Facebook has already started to transform the economics of webscale data centre infrastructure by commissioning in-house designs for servers and even processors, and then open sourcing some of that work to create a broad ecosystem around open, commodity hardware.
That process is harder in networking, where there are far more different pieces of equipment to consider, and it is especially tough in the mobile network, with its exceptional performance patterns and its long history of proprietary, optimized designs from a small, closed and protective circle of vendors and chipmakers.
But the financial pressures on operators, and the shift towards and virtualization, has made Facebook’s approach far more viable, at least in large portions of the network. More and more processes will run in software on off-the-shelf hardware, even though some of the more demanding mobile network functions will still need specialized coprocessors to hit optimal levels.
Facebook’s contribution to blowing apart the traditional supply chain is coming from TIP, a telco version of its own Open Compute Project (OCP), and from a companion effort called OpenCellular, which is aiming to define an open platform for low cost base stations, with a particular emphasis on small cells to push coverage to remote areas. This, of course, is a key goal both for Facebook and Google, which are using a variety of methods to accelerate the pace of bringing high speed access to the unserved millions. They are building their own networks in areas where the established operators are slow to act, developing solutions based on drones, stratellites and new spectrum options such as TV white spaces.
In Facebook’s case, it is also contributing some of its in-house developments to TIP and OpenCellular, to make them available to other vendors and hope to stimulate a wide, open ecosystem. This will make the networks more affordable and encourage other service providers to jump into the business of remote internet access – neither Facebook or even Google, surely, will want to be network operators if they can help it. Even in the case of Google Fiber, in the US, the primary motivation appears to be to kick the carriers into action.
The end goal—controlling a billion more users
In the end, the goal for these web giants is to bring more users online to consume their services and view their customers’ adverts. Google has not gone down the road of open source hardware yet, but does routinely open source key software developments, such as Android, to put itself at the heart of disruptive new mobile or telco platforms. In addition to its attempts to shake up the cost of infrastructure build-out, Facebook is also aiming to disrupt the cost of mobile service delivery with its internet.org initiative, which includes offerings like Facebook Basics – which provides some core apps, including Facebook Messenger and WhatsApp, without counting towards the user’s data allowance.
From internet-org and Facebook Basics to TIP and OpenCellular, operators are signing up with various degrees of enthusiasm or reluctance – but signing up they are, recognizing that, despite the compromises they are having to make to their own power, they cannot afford to have no voice at all in the new platforms, for fear the remnants of their influence are lost to new wireless service providers, from cablecos to OTT players.
So Facebook is establishing itself all along the operator supply chain and service platform, but at the TIP Summit, of course, the focus was on hardware and the new cost base.
“We have been working with partners to successfully test in the market and are excited about the ecosystem that has already come together around Voyager as part of TIP,” Jay Parikh, global head of engineering and infrastructure at Facebook, wrote in a company blog post.
Other announcements at the TIP Summit included plans for the inaugural TIP Ecosystem Acceleration Center, which will be hosted by SK Telecom in South Korea, and is intended to be the first of a series round the world. TIP chairman Alex Jinsung Choi, who is also CTO of SK Telecom, said the centre will incubate local talent and accelerate innovation and product development.
And a new project within TIP is the People and Process group, which aims to address the cultural and organisational changes which come with the shift to commodity hardware and software networking.
The new group, co-chaired by Bell Canada and Facebook, will develop and share cultural and process transformation best practice to improve operators’ key performance indicators. Other members of the working group include Accenture, Agilitrix, Deloitte, HPE, NBN, SK Telecom, Tata, Telefonica and Telstra.
NuRAN is first adopter for OpenCellular
Of even more direct impact on mobile operators is OpenCellular, to which Facebook has already promised to contribute its core design for a rural small cell, for the Terragraph urban cell, and potentially for the Aries Massive MIMO system. Four months after it unveiled OpenCellular, Facebook announced at the TIP Summit that it had now open sourced all the project’s flies, including "schematic, lay-out, CAD files, specs and protocol as part of the TIP Access”.
The speed of progress from Facebook itself has been impressive, but it needs to attract more third party support for commercial products based on its platforms. One small but highly relevant supporter, however, is NuRAN Wireless of Canada, whose LiteRAN 2G base station for remote areas is being validated under TIP, and which will distribute the OpenCellular products as they emerge.
NuRAN provides outdoor GSM small cells with solar power for locations which have no access to traditional telecoms infrastructure. Now it will supply and support OpenCellular offerings as they go into trials. This will start early next year, and Facebook hopes to encourage broad familiarity with its platform to extend the ecosystem.
The TIP and OpenCellular projects are in their very early days, and despite strong early headwinds, it is by no means certain they will achieve the transformations they are targeting in the telecoms industry, rather than becoming bogged down by processes and by the inherent conflicts of interest between many of the members. For MNOs, while they can clearly benefit from low cost, commoditized and open hardware platforms, these also lower the barriers to new entrants, especially as more unlicensed and shared spectrum comes into play, along with radical approaches to wholesale capacity such as bandwidth-on-demand and network slicing.
How open can the mobile network become?
As the network becomes an IT service platform, most of it should be indistinguishable from IT infrastructure, goes the thinking. Cloud computing, and to some extent Wi-Fi, have shown the way to this open infrastructure, not the mobile industry, and the new ecosystems often go hand-in-hand with open source software and, increasingly, hardware – a movement in which the mobile players have remarkably little influence.
A research leader from no more subversive a company than Huawei, Peter Ashwood-Smith, posed the question last year: “Could 5G be more than 90 per cent code from open source origins?”
Ashwood-Smith, head of IP research at Huawei Canada and chair of the ITU’s fixed line 5G focus group, described an entirely open source 4G network which could be deployed today, with Linux-based eNodeB and evolved packet core, Android devices, GNU radios and an open source HSS.
In reality, this approach will take many years to become robust for commercial, large-scale networks. There are limits to the Facebook vision when it comes to cellular networks, especially when it comes to the radios and antennas themselves, which will have to become ever more sophisticated to deliver the promises of 5G, and will remain a source of cost and vendor differentiation for a long time to come. Aries is yet to be tested in anger, but there are sure to be many ways in which the established experts in antennas and MIMO will be able to differentiate themselves against it.
The OpenCellular platform
There is likely to be more impact in the near term from OpenCellular’s core platform, because it targets underserved areas where traditional vendors and operators continue to struggle with the economics of mobile broadband. Also, it provides a low cost, multi-operator small cell – a platform for which the industry is crying out.
The initial OpenCellular design is a shoebox-sized unit which runs open source software and is designed to be mass produced by low cost vendors. It is a modular design which is software-defined and so can support a range of current and future protocols for licensed and unlicensed spectrum. It consists of a general baseband computer (GBC), equivalent to a baseband unit (BBU) in a mobile split architecture.
This handles system functions as virtual network functions, and controls a separate subsystem with the software-defined radio, integrated analog front end, and system-on-chip functions where required, plus the antennas. The SDR and SoC versions support various open source and commercial cellular stacks.
Facebook claims its system is very ruggedized and able to work in extremes of temperature and climate, but also self-installable to reduce cost and time to deployment, especially in remote communities. Its engineering team says it has “designed an innovative mounting solution that can handle high winds, extreme temperatures, and rugged climates in all types of communities around the world. The device can be deployed by a single person and at a range of heights — from a pole only a few feet off the ground to a tall tower or tree.”
The split architecture mirrors the wider moves towards virtualized clusters of small cells, particularly to support indoor coverage and capacity. Small Cell Forum recently published nFAPI, an extension of its FAPI interface, which allows small cells from different vendors to interoperate. nFAPI applies the idea to a centralized baseband unit (BBU) which can control a cluster of radio/antennas (theoretically in a multivendor system). Facebook’s idea is similar, but fully open source.
"Many people might not realize that running their own cellular networks is not only possible but also doesn’t require substantial technical expertise" – Khashif Ali, Facebook
Its platform has been design tested in Facebook’s headquarters in Menlo Park, California and outdoor tests are now in process, while Facebook will work with OEMs and ODMs to ensure wide availability of OpenCellular once it goes commercial.
Terragraph and Aries
Terragraph, like Open Cellular (but more futuristic), is targeted at building low cost cellular and WiFi small cells, but this time in dense urban environments. Complementing this, in a different layer of the RAN, is Project Aries, a concept design for a Massive MIMO macrocell for rural areas.
Terragraph will make heavy use of WiGig, the Wi-Fi-like network for 60 GHz spectrum, and the nodes, which can be mounted on lamp posts or buildings, will use low cost chips and components, says Facebook.
It claims Terragraph has an innovative routing protocol for im-roved collision detection and greater reliability, plus “SDN-like” cloud controllers to handle large numbers of cells efficiently and flexibly, targeting capacity where it is required at any one time. The system can support multi-Gbps links and is IPv6-only. Terra-graph is being trialled in San Jose, California.
The first Project Aries (Antenna Radio Integration for Efficiency in Spectrum), prototype supports 96 antennas and 24 data streams, targeting spectral efficiency of 100bits/sec/Hz. So far, Facebook has demonstrated Aries at 71bps/Hz.
Meanwhile, Facebook continues to experiment with new ways to backhaul its remote cells. It can be assumed that OpenCellular will be able to work as the terrestrial portion of Facebook’s various test networks involving drones or satellites with upload to the aerial device without the need for handsets to communicate directly with them. That would enable commodity cellphones to be used, further lowering costs for consumers and providers.
Facebook engineer Kashif Ali wrote in a Facebook Code post at the time of OpenCellular’s launch in July: “The hardware was designed with simplicity in mind, to encourage people to deploy their own cellular networks. Many people might not realize that running their own cellular networks is not only possible but also doesn’t require substantial technical expertise.”
Web giants target the heart of the network
In that sentence lies the opportunity for a low cost approach to ubiquitous internet access – for things as well as people as the IoT develops – but also the threat to traditional operators and suppliers. The threat to established vendors is obvious, but there are also risks to the MNOs from new competitors which can start from scratch with low cost infrastructure and pursue a new business model.
The web giants are used to making massive hardware investments, and they know how to get their money back on them – still largely through advertising - something mobile operators have yet to figure out, since they are stuck with a direct correlation between the capacity they build and the dollars they can charge for each Gbyte and Mbps of that capacity.
Facebook and Google start from a very different place. At the Facebook F8 developer conference in April, CEO Mark Zuckerberg out-lined a 10-year roadmap to connect the whole world with its services, messages and APIs. This spans technologies at every point in the chain, including apps, robotics and artificial intelligence, and the network itself.
Like Google, Facebook presumably does not want to be a network vendor, a modern version of Ericsson, but is showing the ecosystem what can be done, and hoping OEMs – old or new – will take its ideas and turn them into large-scale commercial systems. “It is not our intent to build and maintain and deploy these networks ourselves. We want to advance the state of the art,” said Jay Parikh, VP of engineering at Facebook.
Copyright © 2016, Wireless Watch
Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.
*DWDM puts data from different sources together on an optical fibre with each signal carried by a different light wavelength. Because up to 80 different wavelengths of light can be transmitted over a single fibre in this way the technique allow more data to be transmitted without laying additional cable, and is thus very attractive to service providers.