A 2016 survey by TMForum, Orchestration: Get Ready for the Platform Revolution, found that "orchestrating services end to end across virtualized and physical infrastructure, including partners' networks, is proving to be one of the most difficult operational challenges for communications service providers and their suppliers." As early as 2014, Axel Clauberg, VP of Aggregation, Transport, IP and Infrastructure Cloud Architecture at Deutsche Telekom, coined the phrase "zoo of orchestrators" to describe the mish-mash of management systems vendors were pushing to manage their siloed NFV solutions.
Orchestration – from zoos to layers
Fast forward to 2018 and these issues have still not been entirely resolved.
Take the mobile domain, for example, where operators may have different suppliers for their RAN and core networks, each with its own preferred orchestration platform. What operators need is orchestration across these sub-domains, not just within silos, in order to maximize efficiency at a global level. Network orchestration is key to managing increasingly complex networks, enabling concepts such as network slicing and VNF (virtual network function) chaining. With orchestrated vRAN, radio resources are coordinated by a centralized network orchestrator for rapid instantiation and automation. NFV with network orchestration of RAN resources will allow services to be turned up in minutes, instead of months, and programmed to automatically scale or self-heal. Service providers can tailor quality of experience for specific consumer and enterprise customers, opening up the potential for new value-added services.
Such capabilities will become increasingly important in 5G networks.
On top of the network orchestration layer (essentially the ETSI MANO concept), operators are also deploying a service orchestration layer that handles operational aspects not covered by MANO. These include translating and decomposing orders that come from a northbound OSS/BSS into mappings that the underlying network orchestrator can understand. Service orchestration also covers fulfillment configuration and coordination of tasks across multiple domains, such as workforce management (e.g. raising a ticket to get a technician dispatched), supply chain management (e.g. getting CPE to the technician), and service activation.
While this two-layered approach of service and network orchestration is an improvement on the zoo of orchestrators that operators faced four years ago, the reality is not quite as simple as two layers. In practice, operators may also need service lifecycle management tools that complement service orchestration and configuration management tools that support their network orchestrators. Network and service management remains a complex field.
The importance of open source
No discussion of orchestration can avoid mentioning open source, given the importance of the Linux Foundation's ONAP project to operators' plans. A recent Heavy Reading report, NFV Orchestration: Evolving Ecosystems & Solution, found that of the 14 orchestration vendors surveyed, 11 were members of ONAP, while just three were members of the ETSI-run Open Source MANO (OSM) project. All three OSM members were also ONAP members, hedging their bets.
The second release (Beijing) of ONAP this June showed good progress around architecture such as the alignment with TMF/MEF APIs and support for microservices and Kubernetes. Beijing is also easier to deploy, lowering the barriers to adoption and increasing the level of automation. Code commits are still coming mainly from AT&T, Huawei, Amdocs, ZTE and Intel, but more operators seem to be getting involved in trials. In addition to the ones that have been vocal ONAP supporters -- AT&T, China Mobile, China Telecom, Orange, Bell Canada -- it seems Verizon and Vodafone are now quite active. And Deutsche Telekom has been submitting use cases and functional requirements for the next release, Casablanca, which is due out this November.
ONAP is by no means the only open source project that operators are engaging with to accelerate innovation and decrease vendor lock-in.
- The Open RAN Alliance (ORAN), founded by AT&T, China Mobile, Deutsche Telekom, NTT DoCoMo and Orange, is working on an alternative to the Common Public Radio Interface (CPRI) between baseband and remote radio units that would allow operators to use a baseband from one vendor with a radio from another.
- The Linux Foundation's Akraino Edge Stack is creating software to support operator-owned clouds at the "edge" i.e. customer premises, basestation or central office. The project will leverage existing open-source cloud project OpenStack, adding middleware, software development kits and APIs to address telco-specific use cases.
- The Linux Foundation is also hosting the Acumos project within its Deep Learning group. Acumos is a framework for artificial intelligence project management that should lower the barriers to entry for telcos and other industries. Acumos also aims to provide a marketplace where academics and commercial data scientists can try to sell specialized machine learning models, customized to address a specific industrial use case. Telcos (and other companies) can then experiment with those crowdsourced models on their own datasets and, if they prove useful, pay to use them in production.
This blog is sponsored by Amdocs.
— James Crawshaw, Senior Analyst, Heavy Reading