Wikipedia Fully Migrates to New Data Center for Better Speed and Reliability

The new data center is now capable of handling all traffic by itself

By on January 21st, 2013 10:38 GMT

Wikipedia is one of the largest websites in the world. While it's not a terribly complex site, in a world where websites are more like apps than like documents, Wikipedia stays true to the initial goal of the web, to make information publicly available and it does this with simple pages, text, images and other media content.

Still, the sheer amount of traffic and the interwoven nature of the site does mean that Wikipedia needs quite a decent data center (or two) to operate.

On a shoestring budget, compared to other major sites, Wikipedia does manage to work properly the vast majority of time for the hundreds of millions of people visiting it each month.

The site, in fact all Wikimedia, the foundation behind Wikipedia, sites are hosted mainly by one data center in Tampa, Florida.

Wikimedia has been working on moving to a new location in Ashburn, Washington. In fact, Wikimedia started deploying servers at a colocation facility there in early 2011, but the migration was slow.

The servers in Ashburn only handled some small tasks at first, but their role grew in time. At this point, the data center there serves most of Wikipedia traffic, 90 percent of it, everything that can be cached.

That left just 10 percent of traffic that needed to actually access the Apache web servers and the MySQL database that powers Wikipedia.

Even so, the old Tampa data center has been crucial as a major outage in August showed, left without connectivity to the MySQL database, the entire site went down even if the Ashburn data center was operational.

Wikimedia has announced that it has now moved all operations to the Ashburn location, including web server and database. At this point, that data center can operate fully independently. The Tampa location will still be used, but mostly as a backup, in the case of an outage at the main site.

Wikimedia also revamped some of its infrastructure in the move, so overall, this should mean better performance for users and better reliability.

Comments