<div dir="ltr">Most of the performance hit is because of commercial actions, not censorship. <div><br></div><div>When there is a tri-opoly, with no opportunity of competition, its easily possible to set prices which are very different than market conditions. This is what is happening here. </div><div><br>Prices are set artificially high, so their interconnection partners wont purchase enough capacity. additionally, the three don't purchase enough to cover demand for their own network. Results in congestion. </div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Mar 2, 2020 at 2:49 PM Pengxiong Zhu <<a href="mailto:pzhu011@ucr.edu">pzhu011@ucr.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">You seem to be implying that you don't believe/can't see the GFW</blockquote><div><div dir="ltr"><br></div><div>No, that's not what I meant. I thought mandatory content filtering at the border means traffic throttling at the border, deliberately or accidentally rate-limiting the traffic, now </div><div>I think he was referring to GFW and the side effect of deep packet inspection. </div><div><br></div><div>In fact, we designed a small experiment to locate the hops with GFW presence, and then try to match them with the bottleneck hops. We found only in 34.45% of the cases, the GFW hops match the bottleneck hops. </div><div dir="ltr"><br>Best,<br>Pengxiong Zhu<br>Department of Computer Science and Engineering<br>University of California, Riverside</div></div><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Mar 2, 2020 at 1:13 PM Matt Corallo <<a href="mailto:nanog@as397444.net" target="_blank">nanog@as397444.net</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">> find out direct evidence of mandatory content filtering at the border<br>
<br>
You seem to be implying that you don't believe/can't see the GFW, which<br>
seems surprising. I've personally had issues with traffic crossing it<br>
getting RST'd (luckily I was fortunate enough to cross through a GFW<br>
instance which was easy to avoid with a simple iptables DROP), but its<br>
also one of the most well-studied bits of opaque internet censorship<br>
gear in the world. I'm not sure how you could possibly miss it.<br>
<br>
Matt<br>
<br>
On 3/2/20 2:55 PM, Pengxiong Zhu wrote:<br>
> Yes, we agree. The poor transnational Internet performance effectively<br>
> puts any foreign business that does not have a physical presence (i.e.,<br>
> servers) in China at a disadvantage.<br>
> The challenge is to find out direct evidence to prove mandatory content<br>
> filtering at the border, if the government is actually doing it.<br>
> <br>
> Best,<br>
> Pengxiong Zhu<br>
> Department of Computer Science and Engineering<br>
> University of California, Riverside<br>
> <br>
> <br>
> On Mon, Mar 2, 2020 at 8:38 AM Matt Corallo <<a href="mailto:nanog@as397444.net" target="_blank">nanog@as397444.net</a><br>
> <mailto:<a href="mailto:nanog@as397444.net" target="_blank">nanog@as397444.net</a>>> wrote:<br>
> <br>
> It also gives local competitors a leg up by helping domestic apps<br>
> perform better simply by being hosted domestically (or making<br>
> foreign players host inside China).<br>
> <br>
>> On Mar 2, 2020, at 11:27, Ben Cannon <<a href="mailto:ben@6by7.net" target="_blank">ben@6by7.net</a><br>
>> <mailto:<a href="mailto:ben@6by7.net" target="_blank">ben@6by7.net</a>>> wrote:<br>
>><br>
>> <br>
>> It’s the Government doing mandatory content filtering at the<br>
>> border. Their hardware is either deliberately or accidentally<br>
>> poor-performing.<br>
>><br>
>> I believe providing limited and throttled external connectivity<br>
>> may be deliberate; think of how that curtails for one thing;<br>
>> streaming video? <br>
>><br>
>> -Ben.<br>
>><br>
>> -Ben Cannon<br>
>> CEO 6x7 Networks & 6x7 Telecom, LLC <br>
>> <a href="mailto:ben@6by7.net" target="_blank">ben@6by7.net</a> <mailto:<a href="mailto:ben@6by7.net" target="_blank">ben@6by7.net</a>><br>
>><br>
>><br>
>><br>
>>> On Mar 1, 2020, at 9:00 PM, Pengxiong Zhu <<a href="mailto:pzhu011@ucr.edu" target="_blank">pzhu011@ucr.edu</a><br>
>>> <mailto:<a href="mailto:pzhu011@ucr.edu" target="_blank">pzhu011@ucr.edu</a>>> wrote:<br>
>>><br>
>>> Hi all,<br>
>>><br>
>>> We are a group of researchers at University of California,<br>
>>> Riverside who have been working on measuring the transnational<br>
>>> network performance (and have previously asked questions on the<br>
>>> mailing list). Our work has now led to a publication in<br>
>>> Sigmetrics 2020 and we are eager to share some<br>
>>> interesting findings. <br>
>>><br>
>>> We find China's transnational networks have extremely poor<br>
>>> performance when accessing foreign sites, where the throughput is<br>
>>> often persistently<br>
>>> low (e.g., for the majority of the daytime). Compared to other<br>
>>> countries we measured including both developed and developing,<br>
>>> China's transnational network performance is among the worst<br>
>>> (comparable and even worse than some African countries).<br>
>>><br>
>>> Measuring from more than 400 pairs of mainland China and foreign<br>
>>> nodes over more than 53 days, our result shows when data<br>
>>> transferring from foreign nodes to China, 79% of measured<br>
>>> connections has throughput lower than the 1Mbps, sometimes it is<br>
>>> even much lower. The slow speed occurs only during certain times<br>
>>> and forms a diurnal pattern that resembles congestion<br>
>>> (irrespective of network protocol and content), please see the<br>
>>> following figure. The diurnal pattern is fairly stable, 80% to<br>
>>> 95% of the transnational connections have a less than 3 hours<br>
>>> standard deviation of the slowdown hours each day over the entire<br>
>>> duration. However, the speed rises up from 1Mbps to 4Mbps in<br>
>>> about half an hour.<br>
>>><br>
>>><br>
>>> We are able to confirm that high packet loss rates and delays are<br>
>>> incurred in the foreign-to-China direction only. Moreover, the<br>
>>> end-to-end loss rate could rise up to 40% during the slow period,<br>
>>> with ~15% on average.<br>
>>><br>
>>> There are a few things noteworthy regarding the phenomenon. First<br>
>>> of all, all traffic types are treated equally, HTTP(S), VPN,<br>
>>> etc., which means it is discriminating or differentiating any<br>
>>> specific kinds of traffic. Second, we found for 71% of<br>
>>> connections, the bottleneck is located inside China (the second<br>
>>> hop after entering China or further), which means that it is<br>
>>> mostly unrelated to the transnational link itself (e.g.,<br>
>>> submarine cable). Yet we never observed any such domestic traffic<br>
>>> slowdowns within China.<br>
>>> Assuming this is due to congestion, it is unclear why the<br>
>>> infrastructures within China that handles transnational traffic<br>
>>> is not even capable to handle the capacity of transnational<br>
>>> links, e.g., submarine cable, which maybe the most expensive<br>
>>> investment themselves.<br>
>>><br>
>>> Here is the link to our paper:<br>
>>> <a href="https://www.cs.ucr.edu/~zhiyunq/pub/sigmetrics20_slowdown.pdf" rel="noreferrer" target="_blank">https://www.cs.ucr.edu/~zhiyunq/pub/sigmetrics20_slowdown.pdf</a><br>
>>><br>
>>> We appreciate any comments or feedback. <br>
>>> -- <br>
>>><br>
>>> Best,<br>
>>> Pengxiong Zhu<br>
>>> Department of Computer Science and Engineering<br>
>>> University of California, Riverside<br>
>><br>
</blockquote></div>
</blockquote></div>