What we use the money for

From Meta, a Wikimedia project coordination wiki

Other languages: Français

The Wikimedia sites include Wikipedia, one of the 100 most visited websites in the world (as tracked by Alexa.com), and numerous other projects in many languages. This means that this site receives a lot of traffic and demands computing power. For this, the primary use of our funds is to purchase the computer equipment needed to support these activities. Soon, for the first time, Wikimedia will also begin taking some financial responsibility for bandwidth usage.

For the details of what we have done with donations from the start of 2004, please see:

We would like to acquire and set up equipment now to handle our anticipated growth and reliability needs through the end of the year. For the coming quarter (1 October 2004 - 31 December 2004) alone we have provisionally budgeted around $50,000 toward the purchase of new servers. Purchases are expected to include additional database, search, web server and cache servers.

So far we do not have enough cash to cover this budget. However, when we needed money for big server purchases in the past, many small donations covered our needs. To contribute or to check the latest account balances, please see our donation page at wikimediafoundation.org.

We've received some queries about server hardware or hosting/bandwidth donations. Yes, we're potentially interested in those if the equipment is within about three years old. The best initial approach is to ask one of the developers in freenode.net IRC channel #mediawiki.

Longer term we are working on how to scale the databases (which of the many options to use). We're using five at the moment, one for writes and reads, two for queries only and two slow ones on Apache web servers solely for last resort backup. For data see:

  • Squid statistics showing total and cache hits.
  • Ganglia cluster stats showing load. The ones which are mostly blue are the web servers, the red/blue mixed are the squid caches (about 60% max is right for max load for them or connect time suffers) and Suda and Ariel are the database servers.
  • Usage statistics showing monthly hits to the servers for the english wikipedia (the largest)

Our growth is pretty simple: when our servers are fast, we grow to use available capacity until we're slow again. There is still no sign of us hitting the limit on demand, so it appears that we'd have no problem finding a larger audience if we had another $50,000-100,000 to spend - our ballpark growth estimates suggest that we'd end up doing that by the end of the year if we stayed fast until then.

One common question: can we use commodity PCs as web servers? We'd like to, but fitting them in the colocation facility is not currently practical. At present we've filled one rack and are just about to start on a second. When a third is needed we'll probably switch to a room for about the same price. At that point, the reduced recurring costs may make this a more useful approach.

If anyone has suggestions on how we use our money, please feel free to make comments on the talk page - we have a dozen or so people on the technical team and more input is always welcome, since we're after the most effective options we can find!