[ndnSIM] can't get Nack when congestion
alexander.afanasyev at ucla.edu
Mon Mar 11 13:07:37 PDT 2013
Sorry for a little bit delayed answer. There is no need to "emulate" IP using ndnSIM. What we did is we just used available IP simulation capabilities (internet module) to simulate IP-based reaction to the same events.
You can check out this repo https://github.com/cawka/ndnSIM-comcom-stateful-fw for paper's scenarios. I haven't yet fully converted our old simulations from the paper to ndnSIM, but for the congestion-related scenarios, we used two completely separated scenarios: one to evaluate NDN using ndnSIM (https://github.com/cawka/ndnSIM-comcom-stateful-fw/blob/master/scenarios/congestion-pop-ndn.cc) and another to evaluate IP using internet module (https://github.com/cawka/ndnSIM-comcom-stateful-fw/blob/master/scenarios/congestion-pop-tcp.cc).
Did I answer your question? If you need code examples for link failure scenarios for IP, I'll try to dig them out (they are not in the repo yet).
On Mar 10, 2013, at 7:33 PM, Xiaoke Jiang <shock.jiang at gmail.com> wrote:
> Hi Alex,
> In your paper "A Case for Stateful Forwarding Plane", you have compared performance of NDN, Path Splicing and IP, eg, the following figure. How do you implement the IP part?
> <Screen Shot 2013-03-11 at 10.33.27 AM.png>
> My Regards,
> Xiaoke Jiang （蒋小可）
> Ph.D Candidate,
> Dept. of Computer Science and Technology,
> Tsinghua University, P. R. China
> On Mar 11, 2013, at 3:15 AM, Alex Afanasyev <alexander.afanasyev at ucla.edu> wrote:
>> Hi Xiaoke,
>> Thank you very much for your commit!
>> I have merged it with some minor corrections. For the later fixes, you can create a new branch (e.g., git checkout -b issue-27 origin/issue-27-hop-count-fix). You can also put "Issue #27" at the beginning of the commit message, this way github will associate commit to the issue.
>> Regarding to you other question. I'm not entirely sure what is your concern. The available forwarding strategies in ndnSIM do not ask routers to do "retransmission" (as Lixia mentioned in another email, routers can merely "retry" if "asked" by NACK or new interest from the consumer). The whole responsibility to retransmit (in case something was lost) is solely end-client (consumer app) responsibility. Did I misunderstand your question?
>> Thanks again,
>> On Mar 10, 2013, at 5:18 AM, Xiaoke Jiang <shock.jiang at gmail.com> wrote:
>>> Hi Alex,
>>> First, I have fix the hop count = -1 (test passed), maybe we can merge this first. I have sent you a pull request on GitHub. I am also try to design two tags.
>>> P,S, Is there any way to make the Forward Strategy works like IP? Because I want to compare NDN with IP, no retransmission on the middle node, I mean the Nack/timeout leads to rechoose another path and leave all the retransmission work to end nodes. And of course, only one path is selected for one prefix.
>>> My Regards,
>>> Xiaoke Jiang （蒋小可）
>>> Ph.D Candidate,
>>> Dept. of Computer Science and Technology,
>>> Tsinghua University, P. R. China
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ndnSIM