ONAP Beijing Release Developer Forum

Regardless of its domains of application, the policy system at the centre of a closed loop system such as ONAP will need to have some very stringent and dissonant quality attributes. As well as ensuring that the correct policy is in the correct place at the correct time, executing policies must preserve state and yet scale and must take account of their (possibly conflicting) context and yet be adaptable. In this talk we discuss the relative importance of these quality attributes when applied to a system such as ONAP. We draw on our work in Policy to discuss how policies that hold state at run time can be scaled, and how a policy system can be engineered to be adaptable even as its state and context changes. We go on to make some suggestion on how the architecture and implementation of the ONAP policy framework can evolve to meet the S3P/Carrier grade requirements