getting very consistent visitors complains about very slow download speed (10-20kbps)
server logs (30d) : bandwidth peaks are below 6% of server max bandwidth
server team contacted, found no hardware problem during their tests
our conclusion is there is a problem on cloudflare end
Can you share the actual domain your visitors are complaining about, so we can have a look at the issue?
In addition, what plan is your domain on?
Please collect the following information from your visitors that are complaining, and share it with us:
Where are you ( … your visitors) seeing these problems?
Please share all the evidence, that this conclusion is based on.
Can you share the actual domain
I’d rather not disclose URL publicly
what plan
free plan, hense why I use the community support (free users can not open a ticket).
collect the following information
Here are some of the ASN of users complaining
3320 (DE)
41164 (NO)
8151 (MX )
5483 (HU)
all of these ASN are residential, nothing wrong to report
share all the evidence
evidence in OP, nothing wrong to report on server side, bandwidth peaks below 6% of max server capacity so the problem is between server and users, which means cloudflare.
I want to know if cloudflare throttles my site bandwith and downgrades users experience. I see no reason why my visitors should struggle because of cloudflare when my server is at most at 6% capacity. I use cloudflare for more speed, not less speed. 
3320 is Deutsche Telekom, and 5483 is Magyar Telekom (Hungarian Telekom), which is a subsidiary of Deutsche Telekom, and will therefore inherit the bad decisions by Deutsche Telekom.
Deutsche Telekom has been known for years, to be taking sub-optimal routing paths.
There is more information about that, … over here.
This Mexican network do not appear to be peering with other networks within Mexico, and seem to be backhauling their traffic to other locations, before they pass it on to other networks, such as e.g. by taking the traffic to the United States.
Their decisions to connect with other networks far away from their users, are causing sub-optimal routing for their customers (at least, in two different regions), where they are taking the traffic towards Dallas, TX, rather than passing it over to Cloudflare locally (e.g. within Mexico).
Telia Norge, f.k.a. GET Norway, …
If you’re outside of Oslo with this ISP, you can steadily add +15 ms to the latency, regardless of which region you’re in.
It also looks like after Telia Norway finally has passed on the traffic to AS1299 (Arelion, f.k.a. Telia International Carrier), that one of their hops are jumping quite a bit in latency, which could be the result of network congestion within their network
I can add that I have personally given up on anything “Telia”-related (including AS1299) decades ago, due to the lack of quality, including, but not limited to hefty packet loss and network congestion.
That doesn’t sound impossible, … but even if so, that doesn’t mean that Cloudflare is to blame.
Not necessarily, it could also very well be with the user’s own ISP, which it seems to be, at least in several of the mentioned examples.
In addition, there could also be something with the connectivity between the ISP of your server, and Cloudflare.
In the events of bad routing, caused by the ISP’s bad decisions, Cloudflare cannot be blamed.
I see allegations in OP, but there isn’t any crystal clear evidence in OP, which puts the blame definitively on Cloudflare.
you should have saved yourself all that trouble, there is absolutely no way a +10ms latency will cause download speed to drop from 100Mbps to 10kbps. I have already got the issue with my own IP, and verified by switching to FTP that the bandwidth cap was HTTP only. there is simply no way the issue is caused by anything but Cloudflare.
what would help is if you’d escalate this to cloudflare and tell them they’d better multiply my bandwidth by 20 in order to match my actual server bandwidth limit.
You cannot compare HTTP (which is passing through Cloudflare) and FTP (which is not passing through Cloudflare) in the way you’re attempting to do, as that will be two completely different network paths, that the two different types of traffic will travel through.
A such comparison still doesn’t show that there is something to solve from Cloudflare’s end.
I won’t be able to escalate silly decisions by individual ISP’s, such as e.g. AS3320 Deutsche Telekom and AS5483 Magyar Telekom, that takes their customer’s traffic to other continents (e.g. European customers towards North America (US)), which gives a typical minimum latency of ~ 100 - 120 ms, or otherwise issues where individual ISP’s are having congestion issues or otherwise backhauling traffic to far away locations, before passing the traffic over to Cloudflare (or the intermediate ISP/carrier before Cloudflare).
As soon as you have shared something bulletproof, which with no doubt will put the blame on Cloudflare (and not on your ISP, or your visitor’s ISP, … or any other third parties), I will be more than happy to attempt to create an escalation request.
It will however be up to you to demonstrate that properly first.
If you can share a contract between you and Cloudflare, where Cloudflare guarantees to match your server’s bandwidth / throughput, and without any restrictions (e.g. disclaimers / liability exceptions), including that Cloudflare takes responsibility for (bad) actions or (bad) decisions made within the networks of third parties, I will be happy to pass that on too.
This topic was automatically closed after 15 days. New replies are no longer allowed.