Reliance Jio buys Radisys to boost its weight in the 5G platform

Reliance Jio’s impact on the mobile industry has, so far, mainly been felt in the shockwaves it has unleashed on its competitors in India. Like Free Mobile in France, it has harnessed WiFi, asset sharing and convergence to reduce the costs of delivering wireless services and undercut rivals. Even more than Free, it has enabled its ultra-low cost tariffs with the efficiencies of a heavily software-driven, greenfield network.

Now it wants to influence its sector in a different way, by helping to drive future platforms and participate actively in open operator-driven initiatives. It has joined several of these, including the Facebook-inspired Telecom Infra Project and the Linux Foundation-hosted ONAP (Open Network Automation Protocol). It has also spoken in the past about open sourcing its internally developed self-organizing network (SON) technology, JioSON.

Its latest move should help fast-track its entry to the inner circle of operators which are defining the next wave of open platforms for converged, software-driven, 5G-enabled networks. It is to pay $1.72 a share (about $75m) for US software company Radisys. This will bring some valuable capabilities in 5G and Internet of Things technology, but most importantly, RJio will pay a bargain price for the leading light in one of the most important open operator-driven projects, CORD (Central Office Re-architected as a Datacenter), which is part of the Open Networking Foundation (ONF) platform.

RJio will pay cash for Radisys, and expects the deal to close by the end of this year. The software vendor, which has a base in Bangalore, India, will retain its Oregon headquarters and a degree of autonomy, said the statements. If this remains the case after the deal is closed, it suggests that RJio sees Radisys as more than an asset for enhancing its inhouse platforms, but rather as a route to influence developments across the operator community.

Akash Ambani, RJio’s director, was clear in his statements that this is about disrupting the competitive landscape and the supply chain, not merely about getting control of some useful technologies. “Reliance and Jio have been disrupting legacy business models and establishing new global benchmarks,” he commented. “This acquisition further accelerates Jio’s global innovation and technology leadership in the areas of 5G, IoT and open source architecture adoption.”

He said Radisys’s engineers would enable RJio to tap into skills and innovation worldwide, and influence global developments, not just those in India. In particular, and with a firm nod to CORD, he highlighted “our work towards software-centric disaggregated networks and platforms, enhancing the value to customers across consumer and enterprise segments”.

Radisys’s roots lie in ATCA blades and other infrastructure, and it expanded into protocol stacks and small cell software when it acquired Continuous Computing in 2011. Its sales in these core areas have fallen over the years and it has been lossmaking. For the full year 2017, the company reported revenues of $133.8m, down 37% year-on-year, and an operating loss of almost $47m, compared with a loss of almost $10m a year earlier.

It has been seeking to rejuvenate its financials by moving into higher value businesses, notably implementation and integration services related to open source platforms like CORD. But this is a slow build, given the very early stage at which most operator deployments of new open platforms are. So there was a real risk that Radisys would run out of time, making the acquisition by an operator – which wants Radisys’s expertise, not its revenues – a strong solution.

Radisys, in turn, will have an anchor user for its services. RJio has used the luxury of its greenfield status to create one of the world’s most automated, software-driven and distributed mobile networks, with heavy reliance on SON and virtualization combined with low cost hardware. That will provide the Radisys platforms with far more real world, large-scale experiences than the proofs of concepts and lab tests that have been the main focus of CORD so far.

The CORD project reflects many of the concerns of forward looking operators, especially how a virtualized telecoms network can support distributed cloud services and edge compute. This is a particular preoccupation of RJio’s, which has worked with Cisco and others to leverage MEC (Multi-access Edge Compute) for optimized video experiences. For instance, the MNO has integrated a mobile content delivery network (CDN) into its 4G network with edge caches in close proximity to the users, delivering content via edge cloudlets to reduce latency. The solution implements 3GPP standard Control/User Plane Separation (CUPS) to provide greater flexibility in the choice of user planes.

In future Jio believes the same architecture will be used for low latency IoT and mission critical services, and deeper involvement in CORD will help to achieve edge compute goals.

CORD includes open reference implementations for residential, enterprise and mobile access based on open source SDN, NFV and cloud technologies. It started life as part of the ON.Lab open source project, which also includes ONOS, and that has subsequently merged with the Open Networking Foundation (ONF).

It is part of the ONF’s mission to create an SDN platform that is specifically optimized for carriers, rather than adapting something which was originally conceived for enterprises, like the more established, and vendor-driven, OpenDaylight. Like many open source telco projects, CORD is hosted by the Linux Foundation.

CORD’s aim is to enable the telco central office to function as a data center in a distributed and virtualized environment, supporting cloud services including the telco’s own virtual network functions (VNFs) and edge compute nodes.

It supports SDN, NFV and elastic cloud services and integrates with other relevant open sources efforts such as OpenStack, ONOS, Docker, Kubernetes and XOS, all running on white box switches and open compute hardware – including Radisys’s own DCEngine – and on merchant silicon.

The CORD community includes service providers such as AT&T, China Unicom, Google, NTT, SK Telecom and Verizon, as well as vendors Ciena, Cisco, Fujitsu, Intel, NEC, Nokia and Samsung.

In 2016, Radisys donated its virtualized EPC (evolved packet core) to the mobile strand of CORD, putting its flag firmly in an increasingly competitive ground. The following year, Mavenir also contributed a vEPC design and the two vendors demonstrated interoperability – a key MNO requirement to adopt any new architecture, and an important driver for many operators to virtualize at all.

In offering the vEPC under an open source licence, Radisys hoped to gain higher exposure for its technology, accelerating adoption, and thereby to monetize it via integration services. Along with Mavenir, it has been addressing one of the three main elements of the mobile CORD framework – disaggregated and virtualized EPC and RAN, and mobile edge services.

Joseph Sulistyo, Radisys’ former senior director of open networking solutions and strategy, said: “The open source driven model is a fundamental element for the 5G roadmap as it enables extraordinary agility for identifying and responding to subscribers’ needs far more quickly than traditional mobility standards. To effectively and productively achieve these objectives for less, we need to empower and integrate EPC with NFV (for virtualization), SDN (for disaggregation), and cloud (distributed ‘anytime, anywhere’ services and continuous DevOps model).”

In 2017, Radisys rebranded itself ‘Open Radisys’ to highlight its work in CORD and also the Facebook-led Open Compute Project (OCP), in which RJio is also active (as well as its companion project, TIP). The company said it was aiming to become a major integration partner for telco data center infrastructure.

In November, it listed its achievements so far related to CORD, which include several operator collaborations that may be very valuable to RJio, to deepen its relationships with other MNOs and propel itself into the heart of the open MNO ecosystem.

Those milestones were:

• A deal with Verizon for Radisys’s FlowEngine Intelligent Traffic Distribution System as the front end for the operator’s SDN service delivery platform.
• The OCP’s adoption of its CG-OpenRack-19 specification, the basis for Radisys’s DCEngine product.
• Collaboration with China Unicom on Mobility CORD implementation for 5G services.
• A Multi-Access CORD proof-of-concept combining wireless and wireline access infrastructure.
• An alliance with Calix to deliver the first commercial Residential CORD platform.

Radisys CEO Brian Bronson said he wanted to be “the Red Hat of telco infrastructure”, adding: “We open sourced our EPC, we open sourced our RAN, we just got approval that our submission is the telco de facto standard for OCP, so the CG-OpenRack-19 is our DCEngine product,. We do have our own products and capabilities but it is really the integration capabilities that we are highlighting moving forward, like CORD.”

He said there was no turning back for the traditional supply chain now operators have a taste of the new open approaches of initiatives like CORD, OCP and TIP. Bronson says Verizon chose its technology because Radisys was open, and because it had come from a telecom background which most server makers lacked. “The original instantiations of DCEngine, which were called TOCP or Telecom Open Compute Platform, were put together to address telecom when everyone was addressing normal enterprise,” he said. “We got engaged with Verizon, they embraced TOCP and DCEngine and bought $100m of our gear.”

The company also offers CellEngine family, which supports disaggregated RAN software and services to help build a bridge from LTE to 5G. Now called MobilityEngine, this will be of high interest to RJio, which already sports a flexible, distributed network including small cells from Airspan, and uses some of Radisys’s software. MobilityEngine is an “open 5G RAN platform and services offering that supports LTE-Advanced network deployments today, while delivering an easy migration path to 5G,” says the firm. The platform is based on Verizon’s 5G Technical Forum’s specifications and is now migrating to 5G New Radio (NR) too.

MobilityEngine will also support emerging 5G services at the network edge as well as network slicing. It can do this within an ETSI MEC framework, or it can be deployed in a CORD  architecture which disaggregates the access from the core network, and the software from the underlying hardware, while supporting open APIs.

For Mavenir, one of the most aggressive of the vEPC providers, it remains to be seen whether it loses a competitor, if RJio mainly keeps Radisys’s implementation for its own use – or gains a more significant one, backed by the Reliance dollars and market weight. Until now, the two companies, despite their collaboration on multivendor CORD, are also rivals, since both have a vEPC and a fledgling vRAN (following Mavenir’s acquisition of C-RAN start-up Ranzure”.

But these specialist vendors know they have the opportunity to break the stranglehold of the big equipment providers, but also that, unless they work in a very different way – and sometimes as partners – they will miss their chance to be a collective David to the Goliaths of Ericsson, Huawei and Nokia.

Mavenir’s SVP of business development, John Baker, said: “Because of our size we can be aggressive with new business models. You’ve seen the earnings statements from Ericsson and Nokia on the decline of the mobile market and, to be honest, we have to agree with them. They have a huge cost base to support. We haven’t got that legacy hardware to support and can focus on virtualization using open market hardware.”

Like other challengers – and the operators which are hoping their supply chains can be shaken up – Baker claims: “We will see certainly Nokia and Ericsson declining and ourselves taking market share. The market is clearly being disrupted, and as a company that has almost 60% of our employees in R&D, we believe we will be in the top three.”

Network slicing will be a key use case for CORD:

Network slicing is perhaps the concept which will most justify 5G’s claim to offer something entirely different to 4G, and therefore be worth the investment. This will be an important objective for M-CORD, and in August 2017, Turkish network vendor Argela provided the first public practical demonstration of how slicing could look in the M-CORD environment – using another vEPC which has been working closely with ON.Lab, from Quortus.

Argela showed a successful proof of concept (PoC) for 5G network slicing on its M-CORD reference platform, the first operational rack outside ON.Lab. It demonstrated dynamically programmable and orchestrated end-to-end network slicing for 5G using its own ProgRAN architecture for virtualized RANs, and its RAN-triggered slicing technologies in the mobile core.

The PoC demonstrated the virtual partitioning of both the RAN and the core to enable network-as-a-service (NWaaS) for a range of use and business cases. Argela’s core network (CN) slicing technology allows the CN elements of each slice to be separated into modules, which can then be centralized in the data center using virtual machines. These modules can be mixed and matched to provide the most suitable resources and topology for each use case and some CN components can be shared across multiple slices using orchestration.

The PoC, says Argela, demonstrate’ how RAN slices can be dynamically mapped to CN slices in real time to orchestrate end-to-end slicing. The orchestrator may map multiple RAN slices to the same CN slice, and the demonstration also included inter-slice handovers.

The demonstration used Cavium’s OCTEON Fusion-M and ThunderX processors, which are being heavily targeted at vRAN; as well as Quortus’s virtualized EPC (evolved packet core). The latter was incorporated into M-CORD, with its open source ONOS controller and XOS orchestrator.