vr000m / rmcat-eval-test Goto Github PK

consolidated by @zaheduzzaman:

Issues:

• How to define test case in a way that the initial starting behavior does not make a difference in the evaluation? For example an algorithm may rampup slowly and the reduction in bottleneck capacity may appear before it reaches a stable state.
• Some of the tests need to run for a longer period of time. For example, RTT fairness. How long? Again the value may depend on how quickly the algorithms reach a stable state.
• Tests for asymmetric link/path capacity.
• Tests should use higher link/path capacity.
• Ted pointed out that the method is not functional under certain RTT. And wanted to verify if different RTTs.
• Do we need a test case where we put explicit impairments to observe the effect?
• Do we need to mention path capacity variation relative to initial capacity?
• Dither the flow start time especially for RTT fairness test case to avoid flow synchronization.
• Details on short lived TCP model.
• Include AQM as an item in potential test case section.

• Test bed/simulator not another bullet point
• Section 4.2, link capacity to path capacity
• Check if the double bottleneck case is covered by the new test cases

02/04/2014

• varun: check 4.1 move to intro
• varun: remove 4.2
• varun: check 4.3
• varun: remove 4.4/TCP model and discuss in relevant section, else email mailing list
• zahed: merge 5.1 and 5.2 ??? -- may be not :)
• zahed: check section 5.

Check out: 6b7defb for fixes.