Quick answer
How do I choose a proxy location or datacenter city?
To choose a proxy location, pick the datacenter city closest to where your target is hosted, or the city whose users you want to see the site as. Distance sets the floor: 1,000 km of fiber costs at least 10 ms per round trip. RackProxies rents dedicated IPv4 in 34 datacenter cities, from $1.29 per IP for 30 days. No KYC, crypto only.
Proxy location rules of thumb
| Speed of light in optical fiber | About 200 km per millisecond, two thirds of its speed in a vacuum |
|---|---|
| Minimum round trip | Distance in km ÷ 100 = milliseconds (there and back) |
| Real round trips | Always higher: cables follow coasts, roads and seabeds, and routers add delay |
| Behind a CDN | The nearest edge server answers, so the origin's city matters less |
| Cities with a page on this site | 34 datacenter cities in 20 countries, plus Singapore and Hong Kong |
| Choosing a city at RackProxies | Choose the country at checkout; a datacenter city can be requested where IPs are online in that city. Availability varies; the checkout shows what is online |
| Geolocation caveat | Datacenter IPs are registered to hosting networks, not to home internet providers. IP geolocation databases can place the same IP in a different city, or even a different country, from the datacenter that announces it |
Should the proxy be near the website or near the users?
Put the proxy near the website when you care about speed, and near the users when you care about what they see. The two goals can point to different cities.
- Speed jobs (monitoring an API, fetching many pages): every request crosses the proxy-to-website path, so a short path saves time on each one.
- Local view jobs (prices, search results, ads, language redirects): the website decides what to show from the country of the IP. Choose the country of the audience, then the closest datacenter city in that country.
- Allowlisted access: the partner only cares that the IP is on its list. Pick a city near the partner's servers, and keep the same IPs at renewal if you can.
When in doubt, rent one IP in each candidate city and measure; that is what the 1 IP plan is for.
How do I find where a website is hosted?
Four checks, all with free tools. Use example.com as a stand-in for your target.
Resolve the hostname
Run
dig +short example.com A(ornslookup example.comon Windows). Write down the IPv4 addresses. Several addresses or a short TTL often mean a CDN or a load balancer.Look up the IP's registry record
Open
https://rdap.org/ip/<ip>in a browser. RDAP (RFC 9082 and RFC 9083) returns JSON with the network name, the holder and a country. The country is where the holder is registered, not proof of where the server stands.Trace the path
Run
traceroute example.com(Linux, macOS) ortracert example.com(Windows). Router names near the end often contain a city or airport code, such as fra, ams or iad. Treat them as hints, not facts.Check for a CDN
If the holder is a CDN or cloud edge network, the address you reached is an edge server near you. The origin can be anywhere; choose by audience instead of by origin.
Why does a CDN change the answer?
A content delivery network answers from the edge server nearest to the visitor, so the "hosting city" of a site behind a CDN is wherever you stand. Many large sites publish one IP that is announced from dozens of cities at once (anycast). Your traffic reaches the closest one.
For a proxy, this means two things. First, almost any datacenter city will reach a CDN edge quickly, because CDNs build edges in the same hubs as datacenters. Second, the content the edge serves depends on the country of your IP. So for sites behind a CDN, choose the country first and the city second.
How much latency does distance add?
Each 100 km of straight-line distance adds at least 1 ms to a round trip, because light in fiber travels about 200 km per millisecond and a round trip covers the distance twice. The table below gives that floor for common pairs.
The distances are great-circle (straight-line) distances computed from city-center coordinates. They are lower bounds: real cables are longer, and every router on the way adds a little. How far above the floor a real path lands depends on the route, so measure your own.
Minimum round trip between datacenter cities
| From | To | Great-circle distance | Minimum round trip in fiber |
|---|---|---|---|
| New York | Ashburn | 350 km | 3.5 ms |
| London | Amsterdam | 360 km | 3.6 ms |
| Amsterdam | Frankfurt | 360 km | 3.6 ms |
| Tokyo | Osaka | 390 km | 3.9 ms |
| Ashburn | Chicago | 910 km | 9.1 ms |
| Tokyo | Seoul | 1,150 km | 11.5 ms |
| Dubai | Mumbai | 1,940 km | 19.4 ms |
| Ashburn | Los Angeles | 3,650 km | 36.5 ms |
| London | New York | 5,570 km | 55.7 ms |
| Singapore | Sydney | 6,310 km | 63.1 ms |
| Miami | São Paulo | 6,570 km | 65.7 ms |
| London | Johannesburg | 9,070 km | 90.7 ms |
| Frankfurt | Singapore | 10,260 km | 102.6 ms |
How do I measure the real latency through a proxy?
Measure it with curl's timing variables, a few times in a row, through each candidate IP. The command below prints the time to reach the proxy, the time until the TLS handshake with the website is done, and the total:
curl -s -o /dev/null -x http://USERNAME:PASSWORD@IP:PORT -w "%{time_connect} %{time_appconnect} %{time_total}\n" https://example.com
Through a proxy, time_connect is the leg from you to the proxy. The gap between time_connect and time_appconnect covers the tunnel setup and the TLS handshake with the website, which is the part the datacenter city changes. Compare the median of five runs, not a single run. The code examples show a timing test in Python.
Which datacenter cities can I compare?
Each city page lists its PeeringDB facilities, internet exchanges and distances to other hubs.
Europe
Amsterdam Netherlands · 64 facilities
Dublin Ireland · 21 facilities
Frankfurt Germany · 53 facilities
London United Kingdom · 82 facilities
Madrid Spain · 31 facilities- Manchester United Kingdom · 20 facilities
Milan Italy · 23 facilities
Paris France · 76 facilities
Stockholm Sweden · 27 facilities
Vienna Austria · 30 facilities
Warsaw Poland · 24 facilities
Zurich Switzerland · 34 facilities
Americas
Ashburn United States · 71 facilities- Atlanta United States · 40 facilities
- Chicago United States · 62 facilities
- Dallas United States · 53 facilities
- Denver United States · 28 facilities
- Los Angeles United States · 43 facilities
- Miami United States · 24 facilities
Montreal Canada · 35 facilities- New York United States · 82 facilities
- Phoenix United States · 25 facilities
São Paulo Brazil · 66 facilities- Seattle United States · 35 facilities
- Silicon Valley United States · 57 facilities
- Toronto Canada · 44 facilities
Asia
Dubai United Arab Emirates · 7 facilities
Mumbai India · 49 facilities
Osaka Japan · 12 facilities
Seoul South Korea · 33 facilities- Tokyo Japan · 38 facilities
Africa
Johannesburg South Africa · 14 facilities
Oceania
MelbourneProxy location questions
Which datacenter city is fastest?
None is fastest for every target. The fastest city is the one with the shortest network path to your target, which usually means the closest one. A Frankfurt IP is quick for a site hosted in Frankfurt and slow for a site hosted in Sydney. Measure through one IP per candidate city before you buy a large plan.
Does the proxy city change the prices or results a site shows?
Often, yes, but mostly through the country, not the city. Many sites choose currency, language and catalog from the country their IP database returns for your IP. Some sites go down to the region or city. Datacenter IPs are registered to hosting networks, not to home internet providers. IP geolocation databases can place the same IP in a different city, or even a different country, from the datacenter that announces it. See why a proxy shows a different city.
Can I choose the exact datacenter city at RackProxies?
Choose the country at checkout; a datacenter city can be requested where IPs are online in that city. Availability varies; the checkout shows what is online. Every city costs the same, from $1.29 for one IP for 30 days. The city of each IP is shown in your account after delivery, next to its port and end date.
Is the RDAP country the location of the server?
No. The country in an RDAP record is the country of the organization that holds the address block. A US company can hold a block used in Europe, and the record will still say US. Use RDAP to learn who runs the network, then traceroute and the holder's own documents to learn where.
How many cities should I spread my proxies across?
As many as your job needs and no more. A monitoring job that checks your service from three regions needs three cities. A price check for one country needs that country only. Spreading IPs over cities that do not serve your goal adds latency without benefit. Plans start at $6.45 for 5 IPs.
Do I need ID to rent a proxy in a given city?
No. No KYC: no ID, no selfie, no documents to sign up, top up or buy. You pay in BTC, USDT, USDC, ETH, SOL, LTC, and you log in with an account number; an email is optional. The crypto payments page explains how a top-up is matched and credited.
Test a city with one IP
1 IP for $1.29. No KYC, paid in crypto.
Sources (5)
- RFC 9082: Registration Data Access Protocol (RDAP) Query Format — IETF (2026-10-19)
- RFC 9083: JSON Responses for the Registration Data Access Protocol (RDAP) — IETF (2026-10-19)
- rdap.org: RDAP redirector — rdap.org (2026-10-19)
- curl man page (--proxy, --write-out) — curl project (2026-10-19)
- PeeringDB — PeeringDB (2026-10-19)