Toronto not Vancouver—fight me on NA East vs West

quinnbizz

Member OP

Joined:: Jun 2024
Posts:: 279
From:: Wellington, NZ

1y ago · edited

#1

I've tested both extensively for a European audience, and IMO Toronto wins for UK/DE routing. Take it with a grain of salt, YMMV.

My methodology: spun up identical boxes on Vultr Toronto and Hetzner Vancouver, ran mtr to London and Frankfurt for 72 hours. Toronto averaged 18ms lower to both. Vancouver's advantage is Asia-Pacific, obviously.

The catch? Both providers might be anycasted through Seattle anyway. I haven't verified the actual fiber paths — https://bgp.tools

...

kenji3

Member

Joined:: Jul 2024
Posts:: 138
From:: Osaka, JP

1y ago · edited

#2

My experience with Asia-Pacific locations suggests that geographic labels do not always match routing reality.

For European traffic, I have observed that Chicago and New York often outperform both Toronto and Vancouver. The transatlantic cables from the northeastern United States are well-established.

I prefer Tokyo or Singapore for my own use cases, so this is only secondhand observation.

conbini > datacenter snacks

sagepat

Member

VM hoarder

Joined:: Aug 2024
Posts:: 57
From:: Bratislava, Slovakia

1y ago · edited

#3

Actually, (the (Vultr (Toronto (node)))) sits in (Cologix (TOR1)) which (peers (directly (with (Tata (and (Cogent)))))), whereas (Hetzner (Ashburn)) is (actually (in (CloudCone (Los Angeles (POP))) (with (extra (latency (injected (by (GeoDNS (routing)))))))).

I run traceroutes. (the (proof (is (in (the (ASN (paths))))))).

Post a reply