Business Services have been a new ambition for MSOs over the past decade. The Service Level Agreements, among others, associated with connectivity services, imply typically customer delivery lead times of weeks, up to months when the MSO must go off-footprint. By contrast, the same business goes to the Cloud Service Provider, and gets compute and storage infrastructure at his disposal in minutes.
These businesses are soon expecting network services to be delivered together with compute and storage, in minutes. SD-WAN delivers on these promises, and drives a need for end-to-end Network Orchestration.
Orchestration has become an overloaded word in the industry. While Network Orchestration and Application Orchestration share a lot of technologies coming from the domains of SDN and NFV, the two approaches serve different objectives. The paper will specifically address Network Orchestration, aiming at launching new revenue generating Business Connectivity services rapidly, with Value-Added Services running in private or public cloud data centers chained-in into the Network Service. MSOs will be able to tap into new revenue streams through the extension of their addressable market going off-footprint, adding new Value-Added Services to their product portfolio, and becoming the trusted middle man in the value chain between Enterprises and public cloud service providers.
The rapid market traction of SD-WAN overlay IP networking, enabled by product suites such as Nuage, fuels the requirements for Network Orchestration: overlay to underlay connectivity, connectivity into Public Cloud, Internet Break-out, flexible onboarding of virtualized Value-Added Services (VAS) such as firewalls, wan optimizers, email filters, and others, all easily customizable for each specific enterprise customer of the MSO.
Various Network Orchestrators, open source and others, have emerged. However, while these all enable writing a Network Service Descriptor (NSD) for a specific service chain use case, including a specific VAS over a purpose-built plug-in, significantly more is needed to deliver Network Services and NSDs rapidly, massively, and reliably. An approach is required that allows to produce NSDs as-a-factory,rather than just ‘executing an NSD’.
The paper will discuss a combination of approaches that together allow to deliver Network Services by MSOs in the most competitive way, and more specifically, how higher automation can be achieved through a ‘best-practice’-based abstraction of enterprise services. A combination of novel Communication and Software technologies are combined to, as examples, automatically generate TOSCA types and plug-ins out of (YAML-based) API specifications; write complex NSDs simply and reliably through substitution and decomposition; and manage a library of NSDs through release upgrades of plugins through an extended suite of Continuous Integration and Continuous Testing