As you may know, the Russian government slows down or blocks many websites for users from Russia. As we have noticed, they can do this for other countries as well. Including Google crawler servers. The technology is roughly as follows - if a packet passes through Russia and several other countries, they terminate the connection in violation of the protocol. The web server does not receive confirmation that the connection is closed and hangs until the timeout expires. In addition, they use botnets to artificially manipulate routing, generating artificial traffic from vulnerable subnets. As a result, the connection between cloudflare and origin servers can takes several seconds. Is it possible to exclude peers that violate protocols and probably many international laws from routing?
It is very difficult to catch such errors. Traffic is not blocked constantly, routes are rebuilt to working peers. But some requests either hang or return a timeout error. It spoils the statistics of the crawler and Google Chrome. Which affects the positions in the search engine.
Using our own autonomous network and peers seems like an obvious solution. But then there is no point in using Cloudflare.
