The customer, one of the leading international telecommunication providers needed a partner with proven domain and software expertise to support automation of their network and public cloud access based on Inmanta’s Intent-Based Service Orchestrator.

The customer has developed a solution that allows their own clients to connect their sites, located worldwide, to a specific public cloud provider (AWS, Azure, GCP) using a point-to-point Ethernet connection.

This Ethernet service is an E-line service where endpoint A is on client’s site and endpoint Z is on service provider’s router co-located with a public cloud provider interconnection point. What was definitely a drawback of the solution, was that all service fulfillment actions on provider’s side were performed manually by their provisioning team which made the whole process of setting up a connection error-prone and lengthy.

What was the challenge?

The challenge was to support both the automated configuration of resources in the provider’s network and  the connectivity provisioning with multiple external Cloud Service Providers. The use case was driven by the commercial launch of a new version of the product for which the end-customer was expected to have a near real-time experience when ordering a new service or modifying an existing one. What was needed in order to achieve it was the lifecycle automation (creating, removing, and updating) of L2 network services. The following cloud providers had to be initially supported:

• Amazon Web Services (AWS), Product: Direct Connect
• Microsoft Azure, Product: ExpressRoute (via the Equinix ECX Fabric API)
• Google Cloud Platform (GCP), Product: Interconnect

Additionally, there were several implementation challenges:

• Variety of API to be integrated – netconf/Yang for physical routers, southbound API with Cloud Service Providers and East West API with Inventory Systems
• Modeling of complex data structure used by APIs – e.g. how to describe complex data trees by YANG and used by NETCONF in partner’s model
• Fitting code interacting with APIs into 3rd party orchestrator southbound components – each API was required to expose CRUD (Create-Read-Update-Delete) interface to manage resources it represented; most complicated interfaces required development of custom mechanism to calculate the difference between current and expected state
• Creating top-level service model – the service model uses a modeling language that describes the intent, i.e., the desired state and behavior (e.g., scalability, performance, and security), and supports arbitrary abstraction levels. Based on this intent, the orchestrator automatically derives the appropriate chain of actions. Other features include modularization, service chaining, dependency management, self-healing, dry-running, and release management. Inmanta is designed to support DevOps principles, allowing rapid development, team collaboration, and continuous delivery
• Testing – fully automated testing suite to support CI/CD