I wrote a script to scrape Contabo's looking glass results from their own site. They don't have one. Tried to automate RIPE Atlas measurements but you need credits.
If anyone has RIPE Atlas credits to burn, measurement ID would settle this fast.
Contabo's "global anycast" is just two cities, same continent?
Eleven years running an ISP and I can tell you "global anycast" from a budget host is like "enterprise-grade" on a $5 VPS. Words lose meaning.
What matters is the SLA and what they commit to in writing. If Contabo's terms don't specify latency targets or PoP performance, you have no standing. Read the fine print before you migrate clients.
SingaporeRep said:
I ran four different upstreams. That's not single-homed.
Your four tests all came from VPS providers. That's still single-homed in practice. Residential eyeballs matter.
mark47 said:
Your four tests all came from VPS providers. That's still single-homed in practice.
JokoNord tested from residential. FyreJun tested from IIJ/NTT. Both show the same 170ms+ to Europe. How many more eyeball networks do you need?
At some point the null hypothesis has to be "Contabo's anycast doesn't work for Asia" rather than "every tester is wrong."
fio, iperf, geekbench. results or gtfo.
Same issue from Mumbai. Following this thread for resolution.