My hosting company, ServInt, posted an interesting article back in June titled “How to Pick a Data Center“. The article discusses the two biggest factors on your site’s latency (how long it takes data to get from the server to the client, aka ping time):
- Physical distance
- Network quality
The article also includes a tool for testing connections to ServInt’s three data centers, the kind of tool your hosting company should also provide. For example, from my location in Pennsylvania, US, I can ping the Washington, D.C. location in 71ms. The LA location has a ping time of 111ms, which is 50% slower, and the Amsterdam location is slightly slower than that, at 123ms. My server happens to be in the DC location, so people in a rough geographic area similar to mine (say, eastern North America) are probably having an experience similar to me. People on the West coast are probably having a similar experience that I have in pinging LA.
The point of the article is that you should use a data center close to where your current, and future, clients are. For example, if I ran a Web site about Hollywood, I’d want to host that near Los Angeles; one on politics: near Washington, D.C. This is probably something many people don’t think about when choosing a hosting company, but you should.
Of course you can balance out the inequities using a Content Delivery Network (CDN), but even there, physical locations matter. I know, for example, that I get a lot of traffic from the eastern and southern Asia, as well as the Pacific, Australia, and New Zealand. Because of that knowledge, I choose to use Amazon Web Services for my CDN, as they have more locations in the southern hemisphere than the alternatives.