Let’s not stumble along the path to autonomous networking

Ulrich Kohn
Boots running on road

The ultimate dream is fully automated, autonomously operated networks. For now, though, the operational reality looks quite different. Today’s communication service providers (CSPs) have to manually plan and build networks, and their field services teams need to go onsite to manually install and configure network elements before services can be activated, frequently only after some test procedures have been successfully run.

What’s more, a lot of those steps have to be replicated with each new customer or even if there’s a change to a parameter of an already activated service. If the customer requests additional security features or a different traffic profile, it’s time for site visits or, at the very least, some manual provisioning.

Sure, today’s management systems do provide automated and semi-automated functions. But those capabilities are predominantly only available with single-vendor solutions in cases where a management system is delivered by the same vendor of the underlying equipment. However, as service providers have learnt the hard way in recent years, monolithic single-vendor networks are slow when it comes to introducing new capabilities and services. CSPs have also suffered from the high cost of adding new capabilities to their networks.

Standardizing disaggregation

There is a clear trend in our industry to disaggregate networks. Separating physical forwarding and transmission from network control or, more generally, any physical network function from the software running on it, allows a service provider to easily select the best technology for each network domain

Disaggregation requires open, standardized vertical as well as horizontal interfaces. There are various standards bodies, industrial alliances and open communities currently addressing this need by developing respective interfaces and protocols. They all start from different technical standpoints but share a common vision of automated processes across vendor and technology domains. It’s great that our industry agrees on a common objective. There is, however, a need to harmonize overlapping activities.

This is where the MEF steps in with its strong track record for taking basic technologies, mapping them against real world requirements, identifying gaps and taking action to close those gaps. This process is anchored in a holistic understanding of service requirements and was spearheaded by the definition of the MEF lifecycle service orchestration (LSO) architecture. Unlike other bodies, MEF takes a top-down approach starting with a holistic understanding of service providers’ services and network requirements, breaking those down into discrete technology domains and assuring the availability of operationally applicable interfaces.

MEF 3.0 services

MEF has successfully established a de facto standard for Carrier Ethernet services and introduced the recently announced MEF 3.0 framework – a cloud-centric LSO architecture defining a hierarchical set of horizontal and vertical interfaces. The Presto NRP API, published as MEF 60, is an intra-provider north-south interface enabling orchestration across multiple technology domains. Through the Presto interface, abstracted network resources are managed in a vendor- and technology-agnostic way. The interface provides an abstract view at the underlying infrastructure and introduces mechanisms for lifecycle management of any MEF-compliant connectivity service.

This interface is northbound of an infrastructure and control management system. It keeps the technology-specific functions separated from the services provided by the network. Exposing abstracted resources to a northbound orchestrator reduces complexity, leading to faster network integration and operational simplicity.

Use of programmable interfaces enables automation of service lifecycle management, which leads to significantly faster service activation and modification times (from weeks to minutes). It improves quality of end-customer experience, and can offer a competitive differentiator to a carrier’s services.

Network diagram for MPLS demo

Next month in Paris, ADVA and Amartus will jointly demonstrate the significant practical value of the recently published MEF 60 Presto NRP interface at MPLS SDN NFV World Congress. This showcase builds on commercially available products and technologies. It’s an essential step towards making agile, assured and orchestrated connectivity services available across vendor and technology domains and turning the dream of fully automated, autonomously operated networks into reality.

Related articles