Automation of service delivery procedures: a systemic approach, beyond configuration scripts

Recent developments in the field of IP networks (including Data Center infrastructures) illustrate operators' growing interest in techniques designed to introduce a high level of automation into the construction and production processes of value-added connectivity services. However, most of these techniques are still in their infancy, as they are essentially limited to providing configuration instructions to the various elements involved in delivering a service, sent following an explicit action reflecting an operator's resource allocation decision. Moreover, these automation techniques are deployed in an environment that is most often based on a logically centralized model, where the computational intelligence responsible for producing the configuration instructions communicated to the elements involved in providing a service is embedded in a piece of equipment often referred to in the literature as a controller, or even an orchestrator. This vision of automation is not new in itself: scripts for automating configuration tasks have been widely used by operators for years. In fact, this type of automation alone does not simplify the production and operation of network resources, nor does it significantly reduce service production times, as such automation processes are restricted to a single stage in the design and life cycle of a service.

New technical steps need to be taken towards more systemic automation, going beyond configuration operations alone.

Indeed, automating the production and operation of network services requires more functional elements, such as :

• support for interfaces (ideally standardized) that enable a network's functional capabilities to be exposed, and capture the needs of services (e.g. in terms of bandwidth, security, robustness, etc.), but also those of neighboring networks likely to contribute to the provision of a service (such as an inter-domain virtual private network service), and of users;

• support for computational intelligence capable of correlating service requests with network capacities at a given moment or over a programmed period;

• support for control loops to ensure that network resources are in line with the objectives set by an operator;

• support for (distributed) mechanisms to increase availability while optimizing exchanges between network resources and controller(s) (in terms of signaling traffic volume, in particular);...