Hi, Min Ye asked me to do a review of draft-ietf-bmwg-sdn-controller-benchmark-meth-07. I first have to say I like the fact that the draft has a "sequence diagram" for each of the test cases in the appendix. This will make it easier to design test software without forgetting small things or misinterpreting the text version of the test. Second, in chapter 5.1.1 you might want to consider that topology discovery can have transient results that might be (accidentally) the correct one. Just because the you get the right topology once it doesn't mean that the algorithm is already in a stable state. I would suggest keeping the "test three times for same result" also for the correct topology. Similar in chapter 5.2.2, it might be good to compare the results of procedure step 3 and 4... see if they have a different of one of more. I am not sure if a test result should be just invalid if the difference is larger or if the difference should recorded in the results. Last, the Table of Content link to chapter 3 and the "section 10" link in chapter 4.1 seem to be broken. Henning Rogge