In a highly anticipated milestone, Facebook crossed the 500-million-user mark this week, as noted in a blog post by co-founder and CEO Mark Zuckerberg. The social network launched a user-generated site called Facebook Stories in honor of the event. And in advance of the festivities, Zuckerberg told a conference in Cannes that he fully expects the network to reach the one-billion-user level at some point over the next few years, calling it “almost a guarantee.”
Certainly, it’s possible Facebook could reach that point — maybe even likely, given the rate at which the network has been growing. At the beginning of 2009, Facebook had just 150 million users, meaning it has more than tripled in size in less than 18 months. If it continues at that rate, it could hit the billion mark within the next two years (some estimate the site is adding somewhere between 20 million and 25 million users a month).
Expanding that quickly, however, is likely to become more and more difficult, if only because operational issues become harder to solve as a company reaches that kind of size. Just look at a network like Twitter — while not exactly the same as Facebook, it operates a similar kind of distributed social network, with about 65 million tweets a day to handle, according to the latest estimates from the company. The last six months have shown how difficult it is for Twitter to maintain even a moderate amount of reliability and uptime; the company has repeatedly been forced to roll back features, throttle or block access and use other means to reduce the strain on its network as it tries to deal with the growing pains associated with having almost 200 million users.
Facebook is even larger, and dramatically so: Users share more than 30 billion pieces of content — including billions of photographs — every month. It’s to the social network giant’s credit it’s managed to avoid similar reliability issues or performance degradation (so far). In part, that’s because Facebook runs a portal where users come to interact, whereas a majority of Twitter’s activity comes in through third-party services and its API.
Lessons From Google
As Om has pointed out, the key to large-scale social networking of the kind Facebook and Twitter are trying to do is infrastructure — the servers, databases and other tools that support those millions of real-time actions and relationships and information. As if to reinforce that point, both Twitter and Facebook are building their own data centers in order to control more of their own destiny, instead of having to rely on “cloud” providers such as Amazon or Rackspace or Joyent. This is one lesson both companies have likely learned from Google, which early on focused on building its own data center infrastructure. As far back as 2006, the search giant had the largest server cluster in the world — larger than NASA or the National Security Agency. One of the company’s biggest fixed costs is power to run all its data warehouses.
The other lesson that Facebook could learn from Google is continual iteration on everything from servers to database formats to load times for webpages. Google-watchers say the web giant continually reviews all of the various points at which its systems are vulnerable, tweaks performance, tests and re-tests, builds in multiple redundancy and so on. In a note posted to an official Facebook blog this week about the growth to 500 million, the company talked about how it approaches dealing with such large numbers from an engineering standpoint. This includes the need to develop and test new features quickly, because “every week we have our biggest day ever.”
Luckily for Facebook, it is now reportedly closing in on $1 billion in annual revenue and has raised several hundred million dollars in financing over the past year or so. So it should have plenty of resources to spend on building the kind of framework that can handle a billion users posting status updates and uploading photos from millions of different locations around the globe.