Implementing Infrastructure-as-Code for Telecom Networks: Challenges and Best Practices for Scalable Service Orchestration

Abstract

Telecommunication networks and cloud infrastructures have evolved independently in technology, operations, and orchestration. Telecom networks remain operationally static with manual intervention mechanisms at service levels. In contrast, Cloud infrastructures introduced notions such as DevOps and Infrastructure-as-Code that surpassed vendor-specific toolsets, allowing the same operations across heterogeneous infrastructural clouds. The telco cloud has outpaced such offerings, taken by hyperscalers and edge service providers creating a split view across infrastructure. Creating a unified view imposes reconciling IaC with Equipment Vendor Specifications and consolidating management of technology driving the realisation of services from HLD to LLD and into provisioning tasks across a network service delivery chain. The service delivery chain, which spans a multi-layer, multi-domain orchestration paradigm, has been adopted recently with the support for the definition of multi-technology service chains. Still during the service deployment phase, a techno-reality ecosystem does not leave room for the actionable realisation of a HLD service net. At the lowest level, Real-time Network Operating System devices or similar, require vendor-specific toolsets that must be reconciled to/from a generic control domain with a rich set of APIs. In the middle layers, the NF hardware can be VM or Container based and provides Vendor L4 monitoring and monitoring that reports on the same stats differently. Communications service providers acknowledged this challenge in capturing and reconciling the operational tech diverging from Telco Service & Equipment Vendor Specifications into the overarching control domain, capturing closing infrastructural gaps, and adopting the technology enabling unified views across cloud and network. Still, CSPs have limited foresight on how to plant these technologies. On one hand, researchers offered theories and frameworks to close the actionable scalability gap by going through the network service rig. Few observations surfaced on how one could kaleidoscope/dissect an intricate network service delivery within a set of viability shifts and reconcile the ViSol with Telco Service Specifications. This paper highlights the challenges one would face while upscaling an INFRA (topology) to an L2-L5 MANO Cloud and shares how the proof-of-concept approach tackled some of it.