How to choose a data center location for your server

Data in fibre travels at roughly two thirds of the speed of light in a vacuum, about 200,000 kilometres per second. The ITU's planning figure for optical fibre is five microseconds per kilometre, so every 100 kilometres of fibre adds about one millisecond to a round trip. Real routes are longer than a straight line on a map, and measurements of real internet paths show round trips several times slower than that theoretical minimum.

Latency also multiplies. Opening a secure web page takes several round trips before the first byte arrives: one for the DNS lookup, one to open the connection, at least one for the TLS handshake and one for the request itself. A server 100 milliseconds away can therefore cost a first-time visitor several hundred milliseconds.

A ping from your office says little about your visitors. Find out where your users are from your analytics, then measure from those regions:

• Use the test IP or looking glass a provider publishes for each location to run ping and mtr from machines in your users' regions, ideally at different times of day.
• Use a public measurement network such as RIPE Atlas to test from many networks at once.
• Compare typical values and how much they vary, not the best single result.

When you order a HyperDC server, the order form shows an estimate of the latency from your location to each data center, so you can compare locations before you choose.

Where your server stands can matter legally as well as technically. Under the EU's GDPR, personal data may leave the European Economic Area only for countries with an adequacy decision or with safeguards such as standard contractual clauses. Turkey's personal data protection law, the KVKK, has its own rules for transfers abroad, including standard contracts that must be notified to the authority. Some sectors and contracts also require data to stay in a particular country.

Choosing a location inside the required region is the simplest way to meet such rules, but it is not the whole picture: backups, monitoring and support access can involve transfers too. Check your obligations with a legal adviser.

A content delivery network caches images, scripts and other static files close to your visitors, which helps a site with a worldwide audience. Dynamic requests, such as a checkout, a search or an API call, still go to your server, so the application and its database should sit close to most of your users and close to each other.

A second location makes sense when downtime would be costly. Common patterns range from regular backups that can be restored in another region to a warm standby that can take over quickly. Each step up costs more and needs regular testing.