Across the R&E community networks are seeing increased demands for flexibility and programmability to support a wide range of experimental use-cases. These range from rapid, ad-hoc configuration changes, the support of emerging technologies such as Quantum, and the need to react quickly to external threats. These requirements cannot be scalably supported by traditional manual management processes. In addition, the increasing complexity of networking infrastructures makes the orchestration of even simple changes across heterogeneous, trans-national networks extremely difficult to coordinate.
Orchestration and automation are no longer just optional enhancements for research and education networks. They are becoming essential components for scalable, reliable, and flexible network service management. Therefore, the development of effective Orchestration, Automation and Virtualisation tools is a core component of future network structures.
To help with this, Global Platform for Labs (GP4L) offers a testbed environment for developing, testing, and refining service provisioning workflows that are reusable across different environments. GP4L is proving to be an ideal environment to explore how orchestration and automation can make complex services both manageable and reusable.
GP4L – a testbed for future networking
GP4L is a large-scale, distributed experimental infrastructure designed to support advanced networking research. Spanning over 9 nodes across Europe and offering over 40 nodes via the GP4L partners, the testbed is equipped with programmable switches, Data Transfer Nodes, and cutting-edge networking capabilities such as P4, DPDK, and eBPF. Hence, GP4L enables experimentation with novel network architectures and technologies, as well as a unique opportunity to validate emerging orchestration practices in a setting that mirrors the complexity of real-world production networks.
To support coordinated experiments and services in such a diverse and geographically dispersed environment, a flexible orchestration and automation stack is being developed as part of the GÉANT (GN5-2) project. The goal is not only to enable efficient provisioning and lifecycle management of services in GP4L, but also to produce reusable workflows that can be adapted by other R&E networks.
Open orchestration
This orchestration framework is built on a modular combination of open-source tools. The team started by implementing an orchestration stack that combines Airflow for workflow logic and scheduling, Ansible for configuration and deployment, and the GÉANT Lightweight Service Orchestrator. The workflows themselves are designed as Airflow DAGs, with each task calling Ansible playbooks driven by Jinja templates. This structure makes it possible to inject GP4L-specific configurations while keeping the core logic reusable across networks.
Maat plays a central role in this ecosystem by acting as the source of truth for network resources and services. It exposes a set of TM Forum compliant APIs built on flexible data models that make it possible to describe services, service components, device capabilities, and network topologies in a uniform, extensible way. These models are designed for compatibility with a variety of tools and services, ensuring that orchestration workflows remain portable and adaptable.
A key use case explored in this effort has been the provisioning of Layer 2 circuits across GP4L nodes. The workflow for this service has been implemented as a series of Airflow tasks, each mapped to Ansible playbooks templated with Jinja. This structure supports per-device customisation while maintaining a common control flow and interface. It allows service descriptions written against the abstract models in Maat to be translated into concrete configuration steps.
Beyond the lab: PIONIER
This orchestration approach has already been validated not only in GP4L, but also in the production PIONIER network, which supports real-world services for PCSS, the Polish NREN. The ability to reuse and adapt these workflows in both experimental and production contexts highlights their robustness and flexibility, as well as the value of designing with portability in mind from the outset.
By focusing on interoperability, modularity, and standards-based design, the workflows developed for GP4L are well-positioned to serve as blueprints for other R&E networks seeking to modernise their service management infrastructure. They offer a practical path to automation that doesn’t require a complete change of existing systems but instead builds on top of them with reusable components and open interfaces.
Looking ahead, the GN5-2 project will continue to expand the orchestration capabilities within GP4L, adding new service types, enhancing lifecycle support, and refining integration practices based on ongoing experimentation and operational feedback. The ultimate objective is to create a shared library of workflows and orchestration tools that can be easily adopted, adapted, and extended across the GÉANT community and beyond.
Find out more about GP4L and Maat
Read the full online magazine here
