Wikipedia 每年运行花费四千万

September 25, 2013

wikipedia 维基百科,刚才看这篇文章的时候,对wikipedia 和 wikimedia 一直没搞明白。wikipedia是特指 wikipedia.org这个维基百科网站。wikimedia 是不单独出现的,Wikimedia Foundation 是一个组织,维基媒体基金会。也就是类似于一个是产品,一个是公司,大概可能就是这个意思。wikipedia上自己解释wikipedia这个词条的意思:Wikipedia is a collaboratively edited, multilingual, free Internet encyclopedia supported by the non-profit Wikimedia Foundation.

Wikipedia is an incredibly(难以置信的,非常的) frugal(节俭的,花费少的) and efficient(有效率的) nonprofit(非盈利的) organization. It serves(为。。。服务) half a billion(半个十亿,五亿) visits per month. But do you know how much it costs to keep this huge(庞大的,巨大的) website running? According to 2012-13 Wikimedia Foundation Annual Plan(年度计划), the expected(预计的) spending is around $40 million. Some may argue(争论,辩论) that Wikipedia is just a site with "just text", why does it cost so much? Oliver Emberton shared his view.


To keep Wikipedia running and serve 500 million visits per month. First, you're going to need hosting. This isn't a service you can pick up(获得) from GoDaddy (this pic is 3 years old but you get the idea):

每个月5亿用户的访问,这样的服务器不是简单得从GoDaddy 能够获得的。GoDaddy 是国外很著名的域名注册商,空间服务商。

Wikipedia 每年运行花费四千万

Notice you have hundreds if not(if not 不然的话) thousands of computers. They're interconnected(联通的,有联系的) in complex and novel(新奇的,异常的) ways - ways that require teams of experts(熟练的,内行的) to design and build.

Now imagine(想象,猜想) that all the above parts are constantly(不断的,时常的) changing, breaking and under attack(攻击). All. The. Time. All your computers will eventually(最终,最后) die. All will need patching(打补丁) to stay secure(安全). This isn't your home laptop. You can't just take down(拿下,卸载) all of Wikipedia to run Windows Update every night. Have you got a plan for that? To create and maintain(维护) one you'll need a team of experts around the world, and such people usually don't work for llama beans.

Now think of your software. Your software has to work in three hundred languages. Have you ever(永久,曾经) tried changing a program that has to work in three hundred languages? It goes a little like this:

"I'll just add this checkbox to our software. It'll only take 5 minutes!"

5 minutes of coding.

"That was easy!"

Except(除。。。之外) now that feature needs testing in 27 different browsers, where you'll learn it doesn't work on Windows 7 Coal-Powered Edition, and you forgot to support pink Blackberry phones and oh, yeah, it needs a team of people to translate into 300 different languages.

Testing is a feat(功绩,壮举) in itself. Something that seems to work fine when 5 people test it can obliterate(消灭,除去) your infrastructure(基础设施,下部构造) when 5 million people are using it. Scaling technology is hard. That's why every big site from Twitter to Quora to Facebook experienced constant(不变的,经常的) downtime in their early years, because the best brains(智力) in the business still agonise(挣扎,烦恼) over how to make websites work when they get that popular.

Of course, we haven't actually(实际上) dealt with(deal with过去时) the business of running Wikipedia itself. A community used by half a billion people needs constant policing(流量监管). It needs quality control, PR, HR, accounts, policies, and hard working people to raise the paltry(不足取的,无价值的) $40m a year it costs to run from cynical(愤世嫉俗的,冷嘲的) folk(民族) who know the price of everything and the value of nothing.

To summarize(总结) with a quote "The devil(魔鬼) is in the details.".

--- EOF ---