recently found this global latency site. So combined with my results for globalatency/meter.net/ipip.net and my friend's M1 globalatency results. Threw the results into gpt and the below conclusion is what I got.
Do note that my VQ is on static IP with no addons .
MR is via wireguard VPN.
M1 is my friend's connection.
Singtel is vanilla.
All tests done over wifi.
Please remember that latency does not equal throughput.
Yes — with the MyRepublic test added, I would
change the four-ISP ranking slightly. MyRepublic comes out surprisingly well despite being handicapped by the WireGuard path.
My ranking if I could use only one ISP
1. ViewQwest — best overall single ISP
2. MyRepublic — best balanced alternative; likely slightly better natively than your VPN test shows
3. M1 — excellent for mainstream/CDN-heavy everyday usage, but less consistent globally
4. Singtel — excellent specialist routing, especially China, but weakest overall as the only WAN
The important distinction is that
M1 may feel faster than MyRepublic for many everyday websites, while
MyRepublic looks better as an all-round global ISP.
Four-way comparison
I matched
447 tests that returned numerical results on all four ISP reports, rather than comparing only their headline averages.
| 447 identical GlobeLatency tests | ViewQwest | MyRepublic via WG | M1 | Singtel |
|---|
| Median | 436 ms | 510 ms | 518 ms | 591 ms |
| Mean | 623 ms | 699 ms | 777 ms | 833 ms |
| 90th percentile | 1,388 ms | 1,415 ms | 1,658 ms | 1,604 ms |
| 95th percentile | 1,926 ms | 2,074 ms | 2,596 ms | 2,237 ms |
| Tests where outright fastest | 221 | 67 | 133 | 26 |
| Overall position | 1st | 2nd | 3rd | 4th |
This is quite revealing. ViewQwest remains comfortably ahead. However,
MyRepublic has a slightly better median than M1 and a substantially better mean and tail latency, despite your MyRepublic results containing the WireGuard detour.
The headline averages from the four reports tell essentially the same story:
ViewQwest 628 ms → MyRepublic 694 ms → M1 787 ms → Singtel 839 ms.
So the ranking isn't being produced by an unusual statistical method.
First, your MyRepublic VPN handicap is quite small
This part of your test setup turned out better than I expected.
From ViewQwest directly to the MyRepublic WireGuard public IP:
- 100/100 packets received
- minimum 7 ms
- maximum 17 ms
- reported average 9 ms.
Through the WireGuard tunnel to its internal IP:
- 100/100 packets received
- minimum 8 ms
- maximum 39 ms
- reported average 10 ms.
And the traceroute is very clean:
ViewQwest → Equinix Singapore → MyRepublic → WireGuard server
with the final hop around 9–10 ms.
That tells us something useful:
WireGuard itself is adding only around 1 ms to the endpoint RTT.
The bigger handicap is the approximately
9–10 ms round trip from your ViewQwest connection to the MyRepublic connection before MyRepublic can send the traffic onwards.
For a simple ICMP ping, I would therefore subtract roughly
10 ms from a MyRepublic-via-WireGuard result to estimate native MyRepublic.
GlobeLatency is more complicated because one HTTPS transaction can require several network round trips — TCP, TLS and HTTP — so the actual WireGuard handicap could easily be
more than 10 ms per complete HTTPS measurement. I would not blindly subtract 30 ms from every result, but somewhere around
10–30 ms of handicap for a fresh HTTPS transaction is plausible.
That means the numbers in the table above are actually somewhat
unfair to MyRepublic.
WireGuard technical reference:
WireGuard
