Wikipedia Part-II : The Cluster
One of the most underreported news is the cluster of Wikipedia servers. Some people might even not consider it news. After all, what’s all that different than other popular websites?
Well, most of us have heard stories like the following, two college students starting a website with one server. Pretty soon the website becomes very popular and they are overwhelmed by the request. Then they starte adding more servers and became multibillion dollar company (Google, Yahoo, etc). But what you don’t hear is there infrastructure after they became successful. Very little is known about the network infrastructure of the top fifty websites and there is a good reason for it too. If you spend millions of dollars building your network, you certainly don’t want unwanted attention from hackers and even worse your competitors. Just to get a rough consultation of cost per server for your network, would run you in thousands of dollars in fees. And building your network at cheap cost is very important step in building your business. Even for me the learning curve was very steep. I started with free server moved to shared hosting then to dedicated server and finally to multiple distributed network. The things you learn on the way can overwhelm you at times. And I am still learning new things everyday.
Well Wikipedia is a whole another story. Probably for the first time we can get a glimpse of the network of a top 50 ranked website. And there are some amazing statistics to ponder about. For a while I thought the top sites require thousands of servers to run. But I was dead wrong. As of September 2005 Wikipedia has only 100 servers. It hosts approximately one million articles, 829082 users at 75 new articles per hour. That’s some impressive statistics. Even more impressive is the fact they did all this with approximately one million dollar/year budget. This kind of real information is a gold mine for webmasters everywhere who dream of making it big in the future. If you think about it, 100 servers is not that many when you are running top 50 ranked website. A million dollar is just a drop in the bucket for the revenue potential of the site. What I don’t understand is Wiki Media Foundation’s refusal to serve advertisement. The site could make lot of revenue by just serving one ad banner per page. They wouldn’t have to rely on donations. Well Yahoo is providing some servers in Asia. I just wonder what Yahoo has up in the sleeves. Maybe a future YPN partner?