The Linux Foundation’s latest project brings together many of the key developments in next generation networks, which the open source organization is taking such a prominent role in shaping. DENT will create an open network operating system (NOS) for disaggregated network switches in campuses or remote offices. This has clear implications for telcos as they build out edge clouds, often in tandem with smaller 5G cells. In doing this, they are targeting enterprise business with on-premise or managed edge/connectivity services, while looking to transform their own network costs with white box and open source architectures.
This is the latest example of the Foundation’s growing role in defining telecoms and enterprise networks, as operators move out of their strictly proprietary worlds into open systems. As its projects proliferate, there is increasing overlap and cross-referencing between them – DENT will no doubt draw on experiences in programs for edge computing, white box switches, NOS and other areas of interest for LF members.
To try to rationalize its activities, and accelerate progress, LF has been gathering some of its projects under umbrella groupings to facilitate cross-fertilization and avoid duplication of effort. These include the Cloud-Native Foundation, the LF Networking Fund, the Deep Learning Foundation and LF Edge.
All these, in various ways, address the convergence of telecoms, especially 5G, and the cloud. They add up to make LF a hugely influential force, via a host of networking and other activities, in helping shape the future telecoms network – cloud-native, open and multivendor, and with distributed networking and data functions converging at the edge.
Though more enterprise-focused than some of the projects, DENT will be very relevant to telcos too, and will offer learnings for other LF activities in adjacent fields. DENT aims to create a common platform for chip providers, systems integrators, equipment vendors and end users, and will focus initially on a specific use case in retail.
Founding members of the group include Amazon, Cumulus Networks, Delta Electronics, Marvell, Mellanox and Wistron NeWeb (WNC) – bringing together white box electronics manufacturers, network silicon providers and cloud companies.
DENT will draw on other Linux-based projects for its core technologies, such as the Linux Kernel and Switchdev, and Amazon will lead the effort to develop the initial seed code. This fits well with the cloud giant’s recent aggressive push into edge computing – last week’s re:Invent conference saw a host of edge-focused announcements, targeting both on-premise services, and those based on managed locations, some housed in telco sites via partnerships with Verizon, Vodafone and others.
The standardized NOS will support campuses, offices, stores and other locations which are remote from the corporate data center, or from the cloud. It will be a simple stack
without abstractions to reduce cost and complexity. The use of Switchdev, a Linux driver for Ethernet switch-chips, will allow developers to treat networking chips like any other hardware, greatly simplifying their task.
Switchdev is valuable because it exposes the innovations silicon to developers without requiring specialist knowledge, enabling system vendors to select the best switch ASICs.
Amit Katz, VP of Ethernet switches at Mellanox Technologies, said: “DENT promotes network disaggregation, which benefits customers by eliminating vendor lock-in, and allows hardware vendors to compete on a level playing field where the very best switch ASICs and systems can win by delivering the highest ROI possible.”
DENT is the latest example of the Linux Foundation’s involvement in shaping edge computing as those platforms emerge. In January, it launched LF Edge, its newest umbrella project grouping, which now has seven projects.
For now, DENT is not under that umbrella but is a high level activity with its own governance structure and board.”
It has remained optional for projects to join one of the Funds. These are designed to establish a common administrative structure to help coordinate and rationalize related projects. In LF Edge’s case, it is described as “an umbrella organization to establish an open, interoperable framework for edge computing independent of hardware, silicon, cloud, or operating system”.
Its first projects were the AT&T-initiated Akraino Edge Stack, EdgeX Foundry (originated by Dell), and the Open Glossary of Edge Computing. They were joined at launch by two brand new projects – Home Edge Project and Project EVE. Later, two more were added – Baetyl, which aims to use container and microservices technology to help developers to build light, secure and scalable edge applications; and Fledge, an open framework for industrial applications.
The broad remit for LF Edge is described as addressing “non-traditional video and connected things that require lower latency, faster processing and mobility”. This will be done via a unified software stack, supporting and drawing on all the involved projects, which does not just cover the telco edge, but also that of the cloud and the private enterprise.
LF Edge aims to tie different edge locations and definitions together to create greater harmonization across industries, and so accelerate deployment of edge devices to reach its projected 20bn figure by the end of next year.
“The market opportunity for LF Edge spans industrial, enterprise and consumer use cases in complex environments that cut across multiple edges and domains,” said Arpit Joshipura, general manager of The Linux Foundation, when the umbrella project was announced.
Meanwhile, another open source initaitive, the Eclipse Foundation, is behind the latest push to ensure that network operators can be full edge cloud providers, despite the interest of the hyperscalers in this market, with a particular focus on industrial use cases.
“We firmly believe that the core infrastructure components of edge computing must be open source” and hosted in a vendor-neutral organization, said Mike Milinkovich, the Foundation’s executive director.
Eclipse has set up the Edge Native Working Group, with eight founders, including two key movers in the industrial IoT, Bosch and Siemens. These two German companies were the first to apply when that country’s regulator decided to earmark spectrum for Industrie 4.0 use, and have advanced programs for harnessing 5G and edge computing to support digital platforms, with a high degree of independence from any specific cloud or telecoms provider.
Other founders include Eurotech, Huawei, Intel, Kynetics, Edgeworx and Adlink.
These will work on a software stack which allows developers to innovate at the edge without having to sign up to a cloud or telecoms operator, or entrust them with code. The Edge Native Working Group is founded on the premise that the new cloud-native enablers, such as Kubernetes, create the opportunity to end vendor lock-in and drive scale and openness into the edge market.
“AWS is sort of like the way Microsoft was in the ‘90s in the sense that it is this huge juggernaut in this new evolving space that nobody can touch,” Milinkovich told SDxCentral. “Lots of people will be very interested in what AWS is offering. Do I think that’s the long term solution for broad adoption of edge computing by a cross section of the industry? Absolutely not.”
The group has released code for its first two projects. Eclipse ioFog, which was contributed by Edgeworx and is being led by its CEO Kilton Hopkins, enables a microservices platform to run on edge devices. Eclipse fogO5, which was contributed by Adlink, offers virtualized infrastructure for fog computing (another term for edge).
“We feel that edge computing is an area that is showing an enormous amount of interest and growth, and particularly with ioFog we’ve got a technology stack here which really helps this exciting Kubernetes ecosystem that’s all the rage at the moment into the edge,” Milinkovich said.
The group will also have a focus on how to implement DevOps processes at the edge (it calls this EdgeOps), because there is “a whole host of very specific and special concerns that pop up when you’re talking about edge that just don’t happen in cloud”.
Edgeworx’s Hopkins is sceptical of the approach that Verizon and other telcos have taken, of partnering with AWS or Azure in order to get edge cloud infrastructure and developer platforms installed quickly and easily in their locations. This can spare the operator the expense of building its own edge cloud, and save it from a head-to-head clash with the webscalers – instead, they get readymade edge capabilities and provide value to the cloud companies with their sites and connectivity.
But Hopkins told SDxCentral: “You have now this need for the carriers to embrace some kind of application infrastructure at the edge and, yes, for some it might seem attractive to grab Amazon. But the thing is, this is the opportunity for the carriers to step away from just being data pipes and actually participate in making and taking some of that value in the world of distributed applications. They couldn’t do it for the cloud. They were the pipe into the cloud but now they’re the ones who own and operate the network that extends to the edge. They have a chance to actually be AWS in those environments if they have the right operating framework.”
To do that, they need to develop their own application layer on top of 5G network infrastructure, and use this to build their own services and developer ecosystem. If this is open and unified, it can be adopted by telcos round the world to form a global base – the lack of global reach is a significant disadvantage for individual telcos when positioning themselves against AWS, Google and Azure. The operators can also tap nito enterprises’ fear of being locked into one cloud provider by offering common connectivity and applications across clouds.
This is the aim of the new Eclipse effort, as well as telco industry-driven initiatives like Ericsson’s Edge Gravity and Deutsche Telekom’s MobiledgeX. These also seek, in different ways, to build an operator applications ecosystem above individual networks and clouds, capable of putting telcos in pole position in many vertical industry value chains. However, the track record of the telecoms operators in creating shared developer and apps platforms is very poor.
Hopkins claims that “open source totally changes that dynamic and I do think that edge computing and 5G put together here are a double threat if an operator does not realize that they need to participate in owning this infrastructure. If they go and they hand all of the money over to Amazon like has been described with the Verizon deal, they’re basically saying: ‘Well, we could have been an application player, we could have been the software player, but we’re going to agree to be a dumb pipe again, another dumb pipe for Amazon’.”