By this stage there should be means to get / set devices’ configuration and state through NETCONF, topology information through BGP LS, network paths through PCEP, etc. The complexity of the domain and vendor specific solutions is abstracted through northbound interfaces of hierarchical SDN controllers. This allows to get information about at least core resources and services in (close to) real time, react on this information dynamically and, what is even more important, automate analysis-reaction loop.
The most noticeable use cases[1] that should be considered for proactive optimization at this phase are:
- Better utilization of resources forcedly overprovisioned long-term to cover major (e.g. yearly) traffic peaks
- Application / service / major customer specific path requirements
- Dynamic BGP peering optimization
- Utilization of network parts not covered by RSVP traffic engineering
- Etc.
Previously it was difficult to automate optimization to such deep level due to complexity of the tasks, overloading of skilled personnel, incomplete non-real-time information about transport network as well as incompleteness of the configuration tools.
[1] Some might be covered by major vendor’s solutions; some require development of custom applications that automate current manual practices