When guessing that it is the website hosts server in specific regions, the strange thing is that they should have some practive with “mass traffic” during specific times - Although the times mentioned, i.e. at weekends in the evening etc. are probably generally still a challenge in specific regions (e.g. Germany is not everywhere as great as one would think regarding modern internet infrastructure).
I think it might well have something to do with he host server (might not be powered well enough or configured properly for the number of people using it or for high-usage periods/ data transmission lines that the website’s server uses might be jammed by a lot of users at particular times of day) - else I would not explain why the problem (at least for me) often only exists for Inaturalist and not for other websites at the same time.
P.S.: Now it is all fine again, the problem really just “pops in” during “Internet Rush Hour”, but not on all websites, only/mainly on Inaturalist.
iNat is a data hog ?
Happens to me since several days. In the cases I checked the details, the image was stored on amazon cloud. I get my internet connection from a reseller via Deutsche Telekom, too.
With often more than 1 MB image size (only downsized by the server to a maximum edge length of 2048 pixels), reviewing a few hundreds of observations (which may have more than 1 image each), that means about 1 GB per identification session…
Same here, Germany, provider: 1&1.
Edit: Fine now, just 20 minutes later.
I think that this happened before on weekends, but usually not this much/often - in the last days it happens more often that I want to identify and stop very soon again due to this issue…
Same here (Germany).
Ping roundtrip time to inaturalist-open-data.s3.amazonaws.com is 166 ms, the delay occurs inside Telekom (using traceroute -T -p 443):
...
3 62.156.246.182 (62.156.246.182) 13.369 ms 13.191 ms 13.175 ms
4 was-sa3-i.WAS.US.NET.DTAG.DE (62.154.5.153) 109.493 ms 109.589 ms 109.703 ms
5 62.159.61.87 (62.159.61.87) 109.273 ms 109.259 ms 110.089 ms
6 * * *
7 52.93.28.116 (52.93.28.116) 110.402 ms * 112.215 ms
...
Whilst I was writing this, the pingtime shrank from 166 to 110 ms.
Experiencing the issues just now. So I decided to start a virtual machine in the Azure Cloud in Ireland. And voilà, everything runs smoothly there.
Has anyone contacted Deutsche Telekom and asked them if they’re throttling https://inaturalist-open-data.s3.amazonaws.com/ ? We don’t think there’s much we can do on our end - in Germany, at least, it seems to be an issue with the internet service provider.
I see a few other people chimed in as well, without providing any information like their location or internet service provider. Again, without information like that, we can’t really investigate or try to find more common threads.
I also changed the title of the topic, as all iNat images are stored on AWS servers.
This might be true, It seems that the Deutsche Telekom was throttling the Internet connections widely, not sure if this is still up-to date… - In this case a workaround for now would be using a VPN
Actually the problem might be in the CDN / DNS configuration of Amazon:
inaturalist-open-data.s3.amazonaws.com currently (slow) gets resolved to s3-w.us-east-1.s3.amazonaws.com, ip 16.182.40.73 instead of a server nearby.
I’m clueless about protocols etc. but I found this (german): https://telekomhilft.telekom.de/t5/Festnetz-Internet/3-Tage-Cloudflare-Probleme/m-p/6498057#M2186792
That was somehow http 3 related, syptoms were similar to what we see.
Unfortunately this is the configuration Amazon makes us use in order to get the subtantial cost savings for hosting the licensed images on that server.
I’m curious if this is still happening for anyone, especially Deutsche Telekom users?
Still very slow on weekdays (afternoon ACT).