Member OP
Hi everyone. Sorry if this is a dumb question but I was wondering how do you actually measure HTTPS latency from lots of locations without spending money? I have like 5 VPS from different providers (OVHcloud, Vultr, and one from Hostinger) and want to see which is fastest for my small website. Thanks!
Ngl this is actually pretty relevant for LLM inference endpoints. I run small models on cheap VPSes and the TTFB variance is wild.
I use a synthetic monitoring tool that hits my API from edge nodes. VRAM isn't the bottleneck here, it's routing. Some of my "budget" boxes in EU outperform the "premium" US ones because of peering.
Member
Windows Server has surprisingly good HTTP.SYS performance for this kind of testing, though I know most of you run Linux. I have not tested against your specific providers. I would be curious to see your results mapped geographically. Routing optimization often matters more than raw datacenter prestige.
Member
Sarah question not dumb!! Many people ask this
I make script with curl and timestamp. Run from сервер in friends apartment, from cheap VPS in 3 country. Network not stable in some place, but data still useful
Server very slow in one location but latency good. Weird!!
Member
I make test in the server with httping and grafana. Actually very easy, friend. How much locations you need? I have friend with VPS in 12 country, he make script for me.
People is surprised when cheap location outperform expensive one. Or what? Routing optimization make big difference, actually. My OVHcloud box in Poland beat Vultr in Germany for Asia traffic.
