That's the mad part. I'm on Hetzner's 20 EUR AX42 in Falkenstein and I get more usable throughput to Dublin than my mate gets from Contabo's 200 to Cork. The pipe is fatter but the route is starved.
He's migrating this weekend. Took three months of "maybe it's my router" before he believed it.
Member
I host for friends in Zagreb who found me at a party, naturally. They all want "something cheap" until their WordPress takes eight seconds to load. I put them on Hetzner Nuremberg because the route to Croatian ISPs is clean. Contabo Düsseldorf? Same distance, worse peering, same price tier.
The party conversation is always the same. "But Contabo gives more gigabytes." "Yes, and I give you a working contact form." They come around.
I compile Chromium for relaxation, so my workflow is embarrassingly parallel and not network-bound. But I keep a Vultr Seoul instance for the occasional binary distribution test. 48 Mbps sustained on 50, every time, because I pay for the tier that actually provisions transit.
Contabo's 100 Mbps would save me money for this use case. It would also waste my time. My relaxation compilation is not relaxing when rsync stalls.
This thread covered CIR, peering, and geography in the first six replies. The last eight are mostly agreement with better anecdotes.
To the person who finds this in 2027: run iperf3 to the actual DC you're considering, from your actual line, at your actual peak hours. The rest is commentary.
Member
Thank you. I will test Vultr Singapore this week and report back if the thread is still open.
