DarioValocchi
commented
7 years ago The use of Mock VIMs to test the API performance of the IA is definitely feasible. We just need an estimation of the actual call time of real deployment, so to embed some kind of wait into the relevant calls of the mock VIM wrappers. A fast measurement of these data can be easily done. Once we have this, we need the SP to be deployed and configured with a topology composed just by mock VIMs and WIM. Some questions arise:
- Is this a separate deployment of the SP?
- Are we changing the SP configuration in every test, or are we relying on the same SP instance to run all the tests?
Stress the IFTA deploying simultaneous NS, try to identify the maximum of simultaneous deployments that the IFTA can perform.