Literature review: Are there any papers that (1) reports issues with inaccurate models (especially models for NFVs or modern software networks), (2) reports traditional testing methods for NFV deployments (people said in industry they're mostly testing with hand-crafted test cases/packets).
Literature about the time/effort needed to build an accurate model.
Batfish(?)'s model vs. inferring from the actual software.