Working it through with Working Group Two

Telco innovation, building a programmable core, deploying on AWS. Here's what wgtwo is trying to do.

As they develop their networks, mobile operators want cloud-native, modern (a vague term, to be sure) software that enables them to operate at lower cost and to innovate new service capabilities more flexibly. Yes… right?

If so, where do they go to get hold of and develop that software? Do they want to enhance their own capabilities and code in-house – say along the lines of the big projects from AT&T, Verizon, or Vodafone’s new in-sourcing strategy? Do they want to lean on their existing suppliers to update their product and ways of working? Do they want to take advantage of new companies that are “native” to cloud operations and software development?

It is, of course, a bit from each column, depending on the sort of each operator we are talking about, their relationships with vendors, the installed base they have, the flexibility they have for change, the target business model they have in mind.

In other words, the “telco transformation” target is shifty as hell, and presents a multi-faceted attack surface for companies trying to innovate in this space.

Erlend Prestgard is CEO of Working Group Two (wgtwo), a provider of core network software for mobile networks. Although wgtwo provides all the functions you would expect to see in a 3GPP core network diagram – “all the TLAs” as Prestgard terms it – he says it is in fact a very different company from most of those in the 4G/5G core space. It builds its software in a different way from most mobile core network providers, and it has some different aims.

The big ambition of WorkingGroup2 is to build the most innovative, programmable core network  technology in the world – to enhance the value of connectivity.

“To do that we needed to radically change the way you operate and deploy core networks… because the current core network is overly complex, expensive and full of legacy.”

“We have deep respect for the telco world, but from a very macro perspective you saw telco dominating the word, then modern tech giants coming in with different operating models and showing 10x innovation capacity. We think we can be a small part of taking back some of that innovation power and making connectivity more useful.”

Perhaps the key words in the above quote are these, “to enhance the value of connectivity – making connectivity more useful.” It’s here where Prestgard targets the gap in telco innovation capability.

“Innovation in telco goes to die at the altar of the cost and complexity of doing things in telcos. And if you do do things successfully it is only one network, getting access to more networks is a nightmare, monetising your effort is a bilateral deal that’s terribly difficult to get. The only thing that comes out of telco innovation is the next G, and big PBX integrations, if I am being brutal about it.

“Twilio, with a 10 year history, owns no towers or fibre and is double the size of Telenor. And, again, to be brutal it built its business on stuff Telenor is giving away for free. So what we take away is that when you make comms assets programmable, easy to work with for developers to integrate into other apps, the value of connectivity increases. That’s the take away for us.”

So much, so familar diagnosis. But what to do about it?

One area Prestgard sees as a barrier to innovation is the way that core networks, and the IMS that drives many operators’ IP service environment, is built.

“You have to think radically for it to change. Take a technical layer – the core network for us – and deploy it across many networks so you get scale. And then step #2 is you API that network so that any developer or operator or end user even can programme the network to suit their needs. And step #3 you have to create a way for developers, and the ecosystem, the appstore… mechanisms for people to get paid and incentivised.”

What wgtwo does is build the core of a core network that presents externally on telco interfaces, but uses different protocols internally. It is also designed to operate on AWS. Prestgard says the aim of that is to make the core product more programmable, and more available to a wider developer base. It also avoids feature bloat within the core.

“So we made a couple of radical choices. We deliver a full core network but in a different way from everybody else. So it’s a full core network – all the TLAs are the same as everyone else – but it is built radically differently.

One key aspect is: I can sell you the whole thing or sell you nothing. I can’t sell you just a box or a gateway.”

The ”buy it all or buy nothing” approach seems to be at odds with the messaging around open core networks, in which we are told that the benefit of the modern, cloud-based core on an SBA is that operators can plug in best of breed software from different vendors, if they want to. Wgtwo seems to be going against that?

Prestgard again: “But that has way more impact than people think. One customer did a survey checking feature requirements and found it ended up using 10% of the features it had asked for in its RFP. This is normal, they pull out the 3GPP spec and list everything that they need, there’s very little critical review. But in reality they only use a tiny fraction of it. So we built the fraction that everyone needs and then we can add the rest as needed. That allows us to build to use cases rather than to a 5,000 page spec document that nobody ever uses.

“Our second principle is that outside of our core, we present to the outside world like any other core. On the inside this doesn’t look like anyone else in the world. The signalling nodes they speak SigTRAN and Diameter on the outside, but on the inside the speak gRPC, modern software protocols.  That means we can take a developer that has never touched telco and they can programme an HLR – our HLR  no longer speaks SS7 or sigttran and our HSS doesn’t understand Diameter, so that allows us to hire different sorts of people, and we believe they can be a lot more efficient on these more modern protocols.”

As Prestgard puts it, “We try to isolate Telco complexity at the edge of the network and then use modern practices inside.”

Additionally, the software is built for the public cloud.

“Everything but the MME and the serving and packet gateway, everything is running in Amazon, in containers, in multi-tenant, and is at least on a journey to API (some is already), microservices, it’s all of that. Not because we like buzzwords but because we built on Amazon to begin with.”

It sounds radical, perhaps too radical, and not really viable for the top level telcos with tens of millions of customers to support.

“The vast majority of partners are not going to buy our service tomorrow. All the central T1 cores – we don’t spend a second pitching to them. We do pitch to them about adjacent cores. Operators that want SaaS and a platform approach, and cost efficiency and innovation power – they are perfect for us. They are the early adopters, or T1s with adjacent use cases, IoT, Private Networks, second brand and SMB and enterprise.”

Why AWS?

“The team started on OpenStack and realised it was spending half its effort on managing the infrastructure. So we stopped that and said let’s not spend time building infrastructure. Let’s move to AWS, and get infrastructure from there. All the functions except EPC are in-house source code from our team. We’re not taking someone else’s product. It’s built by us.”

“The last point is that we deliver as-a-Service. This is hugely underappreciated. In our contract you buy a minimum spec and an SLA, and we commit to a mainstream roadmap development, at no additional cost. So for example, VoNR will be an added feature you turn on, it’s not like you need an RFP for it. We deliver the core network the way AWS delivers compute and storage to us, take the  model and add the application layer towards our customers.”

Deciding to deploy on AWs may create specific dependencies to that environment, but Prestgard still sees the upside.

“How specific we are to AWS, that is a big question, but at least one part is we try to leverage the best of what AWS can give us.

“We have said that it’s impossible for us to compete with incumbents on their terms. Thousands of people over decades have built what they have built. But if we take the best of what Amazon can give us, the most sophisticated security, amazing databases, massive teams that built a lot of services, if you use some of that you get super powers.

“To some extent that is what we have done. Use the super powers of cloud providers to 10x the capacity of the developers. Now there are lots of subtle choices in that. Dependencies you create and the long term strategy is a big question we grapple with all the time. So far the conclusion is at our size you need to stand on the shoulders of a giant.”

So where is wgtwo at, with this vision for a programmable, modern on the inside traditional on the outside, appropriately-scaled core?

It has also been live in Sweden for year and is commercially live and in production in Norway with the primary MVNE in Norway, where it will triggering full migration volumes” any day now”. There’s a base of MNVO customers and it has also put out a public release about work with Hutchison.There are also PoCs in the US with some universities around private campus networks. Italso  provides its core sotware as a service to Mitsui inJapan for private networks – Prestgard says it is in advanced testing and is getting more and more ready to go commercially.

As for competition, there are a number of non-traditional core network providers out here, some aimed at the use cases Prestgard mentions, the likes of Athonet, Druid, Mavenir, Affirmed/Metaswitch/Microsoft as well, how does the competitive environment stack up?

Prestgard’s view: “The industry is splitting its approach into the public core and the private network – creating two product lines that are very different from each other. 

“Our approach is different in that we are building a platform you can scale up or down but it’s the same platform. Everyone else so far, as I understand, is building different products for different use cases. We are inspired by cloud providers – serve the one man business and the biggest enterprises in the world.

“We think core networks can be built generically so you can do the same. You can have a 10 SIM Private Network and 10 million SIM customer base actually using a lot of the same functionality. That is a radically different approach from a lot of others. We think a lot of value will be integrated on top of the network and feel that only way to do that is create a platform anybody can use.”