Thomas Haag, Deutsche Telekom: Access 4.0 – Revolution or Evolution?

Deutsche Telekom has launched the fundamental rearchitected Central Office for fixed access as an answer to operator dilemmas of shrinking revenues and increasing cost. Primary Goals of DT’s A4.0 program are to overcome inflexibility, complex operation and high investment cost. Disaggregation and softwarization of today’s highly integrated vendor specific access appliances and a completely new ecosystem for the network edge are intended to reconcile these challenging goals. This visionary approach is based on community work and therefore open for every operator. In this interview Thomas Haag, Senior Network Architect at Deutsche Telekom Germany, gives insights into the project.

Deutsche Telekom has announced that the first A4.0 deployments in Germany has taken place in january 2021. Could you briefly explain what this is all about?

Thomas Haag: Yes, Deutsche Telekom’s Access 4.0 platform went live. Access 4.0 MVP (Minimum Viable Product) as first step of A4 program comes along with multiple paradigm shifts compared to legacy broadband platforms. First, it’s all about automation. Core of A4 is the A4 POD (Point of delivery). A POD consists of a leaf/spine fabric, compute nodes for control and management purpose and access nodes such as OLTs, DPUs, MSANs. All network components beginning from planning, installing and operation follow the Zero Touch paradigm. That means that processes are clearly decoupled and automized. If an element gets attached to the POD control and management plane the setup is done automatically. All PODs are managed from a geo redundant central management site.

From customer perspective this paradigm applies too. In that case it means if a customer end device like an ONT in FTTH case sends a first sign of life towards the OLT which means in PON case it starts ranging. The POD is processing the port notification and fulfills access configuration and service enabling and policy enforcement in an automated manner. Access 4.0 provides a new platform-based data center technology using merchant silicon based HW so called white box HW, open source software and community work. Which means contribute and benefit from collaboration.

Based on the paradigm described before the architecture is built to be extensible stepwise when new access technologies, new services and new generation of HW comes up. Deutsche Telekom starts with FTTH deployment on the A4 platform in 2020 based on GPON technology. The consecutive steps extend the platform to bitstream access coupling with wholesale and FTTB for copper based inhouse installation combined with PON backhauling. Upcoming steps like integration of DSL is in our roadmap but not planned in 2021.

You have started the initiative in 2016 mainly as a cost improvement program and to overcome vendor locks. Were you able to achieve success?

It has been a long way from starting with AT&T’s CORD initiative to an SDN based access architecture. Yes, we achieved success so far that a community built up among interested operators such like ATT, BT, Telefonica, Turk Telecom and DT support and contribute. From vendor community HW, chipset and switch/OLT manufacturer supporting in a way that we can design and built a platform which support our needs.

Your fundamental approach for the new architecture in central offices is derived from SEBA (SDN enabled broadband access) - created by the Operator Allianz ONF, which Deutsche Telekom is part of. Which experience did you make working within an operator group, where operators may be competitors in the market?

Yes, indeed we are competitors somehow, but we are also partner in community work. This empowers and provides the opportunity to benefit from each other by having a new HW platform, new software platform, new toolset. We shift the level of competition now to another dimension of service level. You mentioned SEBA. This is a perfect example to serve individual needs. If operators are more focused on disaggregate access components like OLT they may choose SEBA 1.0. If other operators may go for a more integrated approach to include also the Service Edge Function (former BNG) under a common control plane with access SEBA 2.0 would be the choice. But in any case, VOLTHA would be the basis for access control and management.

Any operator can use the ONF framework, but not many did so far. What do you think are the most preventing aspects?

All technical concepts correspond also to organization and process changes. This also requires a new set of skills. We call it transformation. Going the path towards softwarized networks requires a transformation inside the companies. So mainly Tier 1 carriers follow in the first step. But I think the acceptance in the market willincrease by seeing results in operation.  

You have made some extensions to the original framework. Can you briefly point them out?

VOLTHA in SEBA stands for virtual OLT hardware abstraction. This primarily focused on PON based access. It started with XGS-PON. Due to DT’s input, specifications on white box OLTs and developing the code are extended towards GPON. Furthermore, the framework now also covers service edge related architecture requirements.

That corresponds also towards dBNG architecture of BBF (TR-459) DT contributed to. Also, OpenBNG and related TIP work-built on merchant Silicon.

Is your approach applicable for other operators?

The aim of access 4.0 is getting productized and being available for the market. This includes operators as well as interested industry partners. Access 4.0 is considered as modular platform serving different architecture flavors.

You are using open source software and standardized hardware components in opposite to SW/HW integrated and validated vendor appliances. What does it mean for you to work with open source software? How does it change your work of working?

By using disaggregated HW and Software changes the entire process of working but keeping the principles.

Before getting live in a network HW and SW needs to be qualified. Qualified means requirements engineering, compliance testing, hardening and bug fixing. But working with different SW and HW partners requires a steering process. Beside partner management a deployment chain for test, integrate, deploy is essential. Now release management becomes part of an operator which formerly a system vendor was in charge off. And in case of using open source SW/components also needs to be part of that process.

For us synergies between global and stable standards and specifications and open source reference implementations is key. This is a little bit in contrast to traditional SDO (Standard Development Organization) approaches.

Getting consensus is now achieved on a different dimension. Being traditional driven based on requirements level we are now driven to achieve consensus on an implementation level. So e.g. ONF is important for us because it is operator driven focusing on operator requirements and use cases.

If your approach were widely adopted, the role of today’s traditional telco vendors would be reshaped. What would you recommend to traditional vendors?

Core competence of traditional vendors is developing products for the telco market. Now vendors move towards system integration business. It consists traditionally of integration of 3rd party in their existing eco system. This is a valuable step but that’s not enough. I recommend moving towards merchant silicon HW platforms and open SW interfaces building a kind of “brick store” enabling operators building or choose their network functions. Finally, I expect that the differentiation will happen on functional level and not on components level.

Can you provide us some insides about the next steps in your program?

Beside maintaining the base system delivering FTTH, FTTB and FTTC is in scope. It is expected to take over beside retail and wholesale also business customers toward A4 PODs. Extending the A4 concept towards other access technologies is not determined and subject for further investigation.

If you would finally judge your 4.0 activity. Is it an evolution or a revolution?

This is difficult to answer. Looking towards market and technology trends it is certainly a logical evolution step towards software defined networking. Considering roles and collaboration with partners it reshapes the eco system which is a kind of revolution.

Thank you very much for this open discussion and good luck with the implementation of A4.0.