Re: [EToSat] [QUIC performance and SATCOM]

Dear Joerg,

If you have any issues in using OpenSAND, do not hesitate to let me know. 

If you are interested in integrate satellite links in the ns-3 QUIC benchmarking framework, you may want to look at SNS-3 : http://www.sns3.org/content/home.php 

Kind regards, 

Nicolas

-----Message d'origine-----
De : EToSat <etosat-bounces@ietf.org> De la part de Joerg Deutschmann
Envoyé : lundi 14 octobre 2019 16:52
À : etosat@ietf.org; Emmanuel Lochin <emmanuel.lochin@isae-supaero.fr>
Objet : Re: [EToSat] [QUIC performance and SATCOM]

Dear Emmanuel,

thank you. I've been aware of OpenSAND for some time, but did not try it yet. The paper certainly is interesting.

When using NetEm, our results were worse than using real links or dummynet. The reason may be that we do not have specific NetEm/tbf settings for satellite links. If anybody has, please tell me. With dummynet the results matched the measurements from real satellite links quite well.

Despite the benefits of OpenSAND, I'm afraid that a lot of developers will stick to the emulation tools they are used to. Therefore, I think it would also be interesting to see how satellite links perform with the ns-3 QUIC benchmarking framework [1].

[1] https://github.com/marten-seemann/quic-network-simulator

Best regards,
Joerg

On 10.10.19 13:52, Emmanuel Lochin wrote:
> Dear Joerg, Tom,
> 
> If you attempt to mimic a satellite link with Netem or Dummynet, you 
> might be interested in having a better emulation system such as 
> OpenSAND
> : http://opensand.org/
> If you need to be convinced by the capabilities of OpenSAND, check 
> this paper : "Making Trustable Satellite Experiments: an Application 
> to a VoIP Scenario" available here : 
> http://oatao.univ-toulouse.fr/21907/
> 
> EL
> 
> Le 10/10/2019 à 12:45, Tom Jones a écrit :
>> On Wed, Oct 09, 2019 at 04:55:58PM +0200, Joerg Deutschmann wrote:
>>> Dear all,
>>>
>>> we continued our QUIC over satellite performance measurements with 
>>> different operators and QUIC implementations. Unfortunately, these 
>>> are still only black-box tests. The results were presented at the 
>>> 25th KaConf. The related paper [1] also gives an introduction to 
>>> QUIC, therefore you probably want to skip to section 4 and 5.
>>>
>>> [1]
>>> www7content.cs.fau.de/~deutschmann/KaConf2019/KaConf2019_QUICoverSat
>>> ellite_paper.pdf
>>>
>> Hi,
>>
>> thanks for sharing these results.
>>
>> Could you expand, maybe off list, the problems you had with netem and 
>> Linux?
>>
>> We have network emulation testbeds here using both dummynet on 
>> FreeBSD and netem on Linux and with some work we managed to get 
>> similar network charactaristics out of both. I do wonder if we might 
>> have missed something that you have seen though.
>>
>>
>>> Best regards,
>>> Joerg
>>>
>>>
>>>
>>>
>>> On 15.02.19 16:48, Joerg Deutschmann wrote:
>>>> Dear all,
>>>>
>>>> thanks for the link to this interesting paper. We did some related 
>>>> work, which you might be interested in. It will be presented at the 
>>>> NetSys
>>>> 2019 conference [1]. A version is available at [2].
>>>>
>>>> While your "QUIC and SATCOM" paper provides more details regarding
>>>> HTTP/2 vs. QUIC, we took a broader view. We measured page load 
>>>> times over HTTP(S)/1.1 and HTTP/2, with and without VPNs. 
>>>> Additionally we set up two Quic servers (Chromium QUIC and 
>>>> quic-go). All tests were run with three different satellite 
>>>> internet providers, showing quite some differences...
>>>>
>>>> Best regards,
>>>> Joerg
>>>>
>>>> [1] www.netsys2019.org
>>>> [2] www7content.cs.fau.de/~deutschmann/NetSys2019_preprint.pdf
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On 09.11.18 04:35, Kuhn Nicolas wrote:
>>>>> Dear all,
>>>>>
>>>>> We have recently posted on ArXiv an article on QUIC and its 
>>>>> performance over a GEO SATCOM link.
>>>>>
>>>>> https://arxiv.org/pdf/1810.04970.pdf
>>>>>
>>>>> This paper has been presented at IETF 103 @MAPRG
>>>>> https://datatracker.ietf.org/meeting/103/materials/agenda-103-mapr
>>>>> g-00
>>>>>
>>>>> The code used to run these experiments is available, so do not 
>>>>> hesitate to contact us if you want to reproduce our experiments.
>>>>>
>>>>> Kind regards,
>>>>>
>>>>> Ludovic Thomas (ISAE-SUPAERO), Emmanuel Dubois (CNES), Nicolas 
>>>>> Kuhn (CNES), Emmanuel Lochin (ISAE-SUPAERO)
>>>>>
>>
>>
>>> _______________________________________________
>>> EToSat mailing list
>>> EToSat@ietf.org
>>> https://www.ietf.org/mailman/listinfo/etosat
>>
> 