Yesterday during the congestion when I was testing the download file on FDC servers, no issues when download from web browser and I can get 10-20MB/s. But if I use curl single stream, sometimes they put me on a congested route and the download suffer at 200KB/s with constant pausing. The buffering on the stream from
@probablye that could be due to the bad paths used during some TCP streams.
At times like this bo bian just tell the upstream not to advertise your prefixes so that traffic can flow from a backup upstream with less problems. It may not be your fault there is a congestion, but end users are still going to demand a certain service standard from you
Go ahead if u think AI slop knows more than some who have insights in the industry and u want to normalise poor traffic engineering

lots of misinformation from above I dunno where to start.