Round trip, 20 packets
ping -c 20 lg.vpsoffshore.comThe median matters, not the minimum. A single fast packet says nothing about a link.
Test the route before you pay for it
Targets, files and the exact commands. No account, no form, nothing to accept — measure our network from your machine and decide on your own numbers.
Targets
Point something at these
| Family | Target | Good for |
|---|---|---|
| IPv4 | 198.51.100.24 |
ping · traceroute · mtr |
| IPv6 | 2001:db8:4a::24 |
ping6 · traceroute6 |
| Host | lg.vpsoffshore.com |
resolves to both, use it in scripts |
Test files
And something to pull
| Size | URL | Use |
|---|---|---|
| 100 MB | https://lg.vpsoffshore.com/100mb.bin |
Enough to fill a gigabit link for a second. Use this one first. |
| 1 GB | https://lg.vpsoffshore.com/1gb.bin |
For 10 Gbps ports and for anything you intend to publish. |
| 10 GB | https://lg.vpsoffshore.com/10gb.bin |
Sustained-throughput testing. Please do not loop it. |
The files are incompressible random data, so a proxy or a CDN on your side cannot flatter the result. They are served from the same port your server would get.
Four commands
What to run, and how to read it
Most “the network is slow” reports are a measurement artefact. These four, read properly, tell you what is actually happening.
The path, hop by hop
mtr -rwzbc 100 lg.vpsoffshore.comLoss in the middle of a path is usually a router deprioritising ICMP, not a broken hop. Loss at the last hop is real.
Throughput, single stream
curl -o /dev/null https://lg.vpsoffshore.com/100mb.binOne TCP stream over a long path is limited by the window, not the link. Expect less than the port.
Throughput, honestly
aria2c -x8 -s8 https://lg.vpsoffshore.com/1gb.binEight streams is what a real workload looks like. This is the number to compare against another provider.
Reading it
Three results that look bad and are not
We would rather you understood the output than believed our figures. A provider explaining how to interpret a bad-looking result against itself is telling you something a marketing page cannot.
Loss in the middle of an mtr
Routers deprioritise ICMP addressed to themselves. If the last hop shows 0% loss, the path is clean no matter what hop 6 says.
A single curl at 40 MB/s
One TCP stream over 35 ms is window-limited long before it is link-limited. Run the eight-stream command before concluding anything.
A first packet far above the median
ARP, DNS and a cold path. It is why the ping command asks for twenty and why we publish a median rather than a best run.
Not yet
There is no interactive console on this page, on purpose
A looking glass that runs commands from our side is being built. Until it answers, we would rather say so than show you a form that does nothing.
Read how to run oneCan you run a traceroute back to me?
Yes — send support your address and we will paste the output from our edge, usually within minutes. The return path is often the interesting one, and it is the half you cannot measure from where you are sitting.
Can I see your BGP table or AS-path?
Ask and we will show you the routes for a prefix you care about. We do not publish the whole table because it names our carriers, and naming them invites pressure on companies that did not choose our positioning — the same reason we do not name the building on the facility page.
My numbers disagree with your latency map.
Then send them. If you are right we will change the published figure, and say why. A page of measurements is only worth anything if it can be corrected by a reader.