A VPS in the wrong city is worse than no VPS

[Fairman]

A VPS in the wrong city is worse than no VPS

Plenty of traders rent a virtual server and assume speed is solved. Then the latency to the broker turns out to be 90 milliseconds because the server sits in a data centre on the other side of an ocean.

The rule is boring: put the machine as physically close to the broker's trade server as possible. Many brokers run their servers in London LD4, New York NY4, or Tokyo TY3. A VPS in the same building can reach the matching engine in one or two milliseconds.

Before paying for anything, ask the broker where the trading server is hosted, then test ping from candidate VPS providers. A cheap server in the right place beats an expensive one in the wrong place every time.

Another detail: ping to a web address is not the same as latency to the trading server. The platform should be connected and the actual round trip time measured from within it, which most terminals can show.

A rough benchmark, for reference: under 2 milliseconds suggests the machine is in the same facility as the broker's server, 5 to 20 suggests the same city, and above 50 suggests a different region altogether. A scalper working at a one or two pip scale will feel the difference between the first and the last.

It is also worth asking the VPS provider about the network route. A cheap provider with a congested path can show a low average ping with occasional spikes, which brings the jitter problem back by another door.

Providers also differ in how they handle outages. Reading recent reviews about downtime and support responsiveness is worth the time, since a cheap server that goes dark during news costs far more than a better one.

What latency does the setup show to the actual trade server?

[Amanhasnoname]

When I first rented a server I picked the cheapest one with good reviews, which happened to be in Germany, while my broker's trade server was in London. Inside the terminal the round trip showed about thirty milliseconds, only slightly better than my home line, and the fills on news were no better at all. I had paid for convenience, not speed.

I asked the broker where their server was hosted, which took two emails, and then moved to a provider in the same London facility. The terminal latency dropped to under two milliseconds. The bigger improvement was consistency: the readings barely moved even during busy releases, where my old setup had jumped around.

One tip for anyone testing providers is to leave a demo terminal running on a trial machine for a full week and log the latency the platform reports. A single reading on a quiet afternoon tells you very little about how the server behaves during the London open.

[FXS HongKong Scalper]

Many brokers run their servers in London LD4, New York NY4, or Tokyo TY3.

Boring but true. For yen and Aussie traders the twist is that some brokers run separate servers for Asian clients, often in Tokyo or Singapore, while others route everyone to London or New York. Asking the broker which data centre hosts the trade server before renting anything saves a month of wondering why the VPS is slow.