[ndnSIM] initial RTO value
Alex Afanasyev
alexander.afanasyev at ucla.edu
Thu May 30 09:58:17 PDT 2013
Hi huyao,
If you're referring to RTT estimate maintained for each FIB entry (https://github.com/NDN-Routing/ndnSIM/blob/master/model/fib/ndn-fib-entry.h#L180), then it is actually not used right now. Again, this is just an artifact of the current implementation in ndnSIM, but routers don't really have a retry timer and don't do anything until a retransmitted interest from an end-user is received or the PIT entry is timed out.
---
Alex
On May 29, 2013, at 6:09 PM, yao hu <huyao0107 at gmail.com> wrote:
> Hi Alex,
>
> Thanks for explanation. I understand. But every intermediate router also maintains the estimate of the RTT for each interface to retrieve data, right? This RTT estimate is used in setting up a retry-timer. Then what happens if the timer expires and the router does NOT receive a NACK? It still does not retry alternative faces?
>
> Regards,
> huyao
>
>
>
> 2013/5/30 Alex Afanasyev <alexander.afanasyev at ucla.edu>
> Hi huyao,
>
> This RTO refers only to retry for end-user retransmission. Intermediate nodes, at least in the currently implemented strategies in ndnSIM, don't retry alternative faces, unless they receive a retransmitted interest from the end-user (retransmitted interest = interest for which there is an existing PIT entry, and interest is received from the face, from which the node already received the interest for the same name before).
>
> ---
> Alex
>
> On May 29, 2013, at 7:38 AM, yao hu <huyao at goto.info.waseda.ac.jp> wrote:
>
>> Hi Alex and Yaogong,
>>
>> Thanks for your both explanations. For my understanding, this RTO is used in setting up a retry-timer. But I am a little confused by the word "retry" in "retry-timer". This "retry" refers to the retransmission for end users for sure, but it also refers to the retry for intermediate nodes (retry other alternative faces than failed one)?
>>
>> Regards,
>> huyao
>>
>>
>>
>> 2013/5/29 Alex Afanasyev <alexander.afanasyev at ucla.edu>
>> Hi Yaogong,
>>
>> Thanks for opening an issue and finding the problem! I just incorporated your commit that fixes the incorrect initialization of attributes.
>>
>> As of now, the initial RTO is 1 second, and later cannot become smaller than the minimum, which is by default set to 200ms.
>>
>> ---
>> Alex
>>
>> On May 28, 2013, at 8:21 AM, Yaogong Wang <ywang15 at ncsu.edu> wrote:
>>
>>> Hi Alex,
>>>
>>> I believe the initial RTO should not be 200ms. According to the code, RTO is set to m_currentEstimatedRtt + 4 * m_variance bounded by minRto and maxRto. By default, m_currentEstimatedRtt is initially set to 1s (InitialEstimation) and m_variance to 0. Hence, the initial RTO should be 1s which is within minRto=200ms and maxRto=200s.
>>>
>>> The current code does set initial RTO to 200ms due to a bug that sets m_currentEstimatedRtt to 0 initially. I have reported the bug here: https://github.com/NDN-Routing/ndnSIM/issues/40
>>>
>>> Thanks,
>>> Yaogong
>>>
>>>
>>> On Wed, May 22, 2013 at 4:40 PM, Alex Afanasyev <alexander.afanasyev at ucla.edu> wrote:
>>> Hi huyao,
>>>
>>> Yes. There is an initial RTO value and it is 200ms. You can control this value by setting "ns3::ndn::RttEstimator::MinRTO" value. ndnSIM uses a modified version of RTT estimator, originally in Internet module: https://github.com/NDN-Routing/ndnSIM/blob/master/utils/ndn-rtt-estimator.cc.
>>>
>>> --
>>> Alex
>>>
>>> On May 22, 2013, at 10:52 AM, yao hu <huyao0107 at gmail.com> wrote:
>>>
>>>> Hi Alex,
>>>>
>>>> Is there a default initial RTO value for consumer node? RTO should be calculated based on the observed RTT samples, but how about the first Interest timeout for some reason? If there is a default initial value, is it 200ms?
>>>>
>>>> Thanks for your explanation~
>>>>
>>>> Regards,
>>>> huyao
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> ndnSIM mailing list
>>>> ndnSIM at lists.cs.ucla.edu
>>>> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
>>>
>>>
>>> _______________________________________________
>>> ndnSIM mailing list
>>> ndnSIM at lists.cs.ucla.edu
>>> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
>>>
>>>
>>>
>>>
>>> --
>>> ============================
>>> Yaogong Wang, PhD Candidate
>>> Department of Computer Science
>>> North Carolina State University
>>> http://www4.ncsu.edu/~ywang15/
>>> ============================
>>
>>
>>
>>
>> --
>> **************************************************
>> 早稲田大学 基幹理工学研究科 情報理工学専攻
>> 後藤滋樹研究室
>> 胡 曜 (HU Yao)
>> E-mail : huyao at goto.info.waseda.ac.jp
>> **************************************************
>
>
> _______________________________________________
> ndnSIM mailing list
> ndnSIM at lists.cs.ucla.edu
> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20130530/9d9b71a4/attachment.html>
More information about the ndnSIM
mailing list