In this final blog post about the TM Forum report on Orchestration, we look at the role of standards and open source in end-to-end orchestration, as well as implementation strategies.
Standards and Open Source: the Rise of Three Camps.
Among standards bodies and open source groups, TM Forum is relied upon by a large majority of service providers for help with orchestration, with ETSI, MEF, and OASIS also cooperating to develop the Hybrid Network Management Platform.
Almost two-thirds of service providers view open source as either extremely or very important for NFV and SDN deployment. Strong alignment is needed in a digital ecosystem of partners where it is important that everyone understands the requirements in the same way and work together on a common source code.
AT&T, China Mobile, and Telefónica have are vying for open source leadership. By contributing ECOMP to open source, AT&T is clearly pushing for its platform to become the de facto industry standard for NFV orchestration. The company is planning to contribute the core orchestration code from ECOMP, not policy or analytics which it considers proprietary. China Mobile, which supports the OPEN-Orchestrator Project (OPEN-O), is unlikely to adopt AT&T’s ECOMP, although AT&T has said that the ECOMP code itself is vendor-neutral and the company will consider other integrators. Telefónica has contributed its virtual infrastructure manager and orchestrator to Open Source MANO (OSM), an ETSI-sponsored group. Ultimately, all the players need to realize that change will come faster if everyone works together and agree on how to federate disparate approaches.
Strategies for Implementing Orchestration
Service providers can move toward becoming platform providers by taking an enlightened strategy toward adopting orchestration. Key elements include:
Understand what end-to-end orchestration means.
Orchestration goes beyond network functions virtualization (NFV). It is about automation. The NFV Orchestrator role specified in ETSI’s NFV MANO isn’t enough. To manage hybrid networks and give customers the ability to control their own services end-to-end automation is required, and that includes operational and business support systems (OSS/BSS).
Adopt a platform approach.
Platform providers like Airbnb, Amazon, Google, Netflix, and Uber have achieved success by providing an interface between customers and sellers. Telecom companies like BT, Orange, and Vodafone see orchestration as a strategic step toward becoming platform providers for third parties, building their businesses by curating ecosystems that link end customers or users with producers of goods and/or services. Network operators need a similar model to offer the network platform as a service.
Determine where orchestration has to happen.
Orchestration happens everywhere, and systems must communicate with each other and with many other physical and virtual elements to deliver a service request that the customer initiates through the customer portal. This spans the technology layer, which includes physical and virtual functions, the resource layer where functions are modeled as logical resources, the services layer where provisioning, configuration and assurance happen, and the customer layer.
Use common information models, open APIs and intent-based management.
A service provider’s master service orchestrator will never have complete visibility into other providers’ networks and operational and business support systems. Service providers will automate service provisioning and management end to end by agreeing to use the same information, data models, and APIs so that orchestrators in different domains can communicate. Intent-based management abstracts the complexity of the network and uses customer intent and policy to manage it. The answer lies less in the orchestrator and more in standardizing the things that are being orchestrated.
Implement closed control loops, policy and analytics.
Closing the loop means collecting and analyzing performance data to figure out how the network can be optimized and then applying policy, usually through orchestration, to make the changes in an automated way.
Design in security.
Trying to bolt security features on afterwards doesn’t work. Detecting configuration-related vulnerabilities requires an orchestrator that can call on internal or external security functions and apply security policies to users or systems accessing NFV components.
Chart the migration paths.
For service providers, success will depend greatly on how well they plan the transition, setting a clear migration strategy both technologically and culturally. This really comes down to learning to think like a software company.
Work toward a common goal in open source groups.
Aligning around a single approach would certainly make end-to-end orchestration easier, but short of that ideal, ways must be found to federate the approaches through collaborative work on common information and data models and APIs. Developing the technology and business models needed in the world of 5G and the Internet of Everything can only happen if everyone works together.
This is the final blog in our series on the TM Forum report on Orchestration. The report confirms that orchestrating services end-to-end across virtualized and physical infrastructure is indeed a huge challenge—but not an insurmountable one.