Weekly Update

Social Networks Need to Grin and Bear Infrastructure Costs

The good news for popular social networking sites like Facebook and Twitter is that they keep on growing and keep on raising more money. The bad news is that continued growth means more money pumped into their infrastructures just to keep up. The real irony here is that, while social networks are cutting-edge web-based services, there is little symbiosis with the other cutting-edge web-based paradigm getting lots of attention lately: cloud computing. Social network sites can’t look to the cloud to help them solve their infrastructural woes.

Just look at Facebook, which, it is estimated, spends between $20 million and $25 million a year on running data centers alone, not to mention the cost of procuring the hardware to fill them. It has more than 1,000 servers dedicated solely to its memcached tier. Facebook plans to add even more data center space later this year. How much of its additional $200 million will go toward infrastructure capital and operating expenditure?

Twitter has been spending the majority of its resources managing growth, and it expects that, by year’s end, most of its employees will just be helping to keep the lights on. How much more can Twitter expect to spend? If its 1,382 percent year-over-year growth is any indication, the answer is “a whole lot.” (Of course, if Twitter doesn’t want to keep playing this procurement game, Yahoo will gladly relieve it of the responsibility.)

And don’t forget the case of social network pioneer MySpace. Although its popularity might be waning, MySpace still generates 2-3 terabytes of new data every day, and it requires a massive data infrastructure in order to store and process this load. While I don’t have any cost numbers on hand, I have to assume MySpace spends a fortune to keep it up and running.

Typically, when companies need to focus their time and money on their core competencies, on innovation — and not on infrastructure woes — its time to turn to the cloud. But for sites like Facebook, Twitter or MySpace, that’s not an option: There is no evidence of which I’m aware that a public cloud could handle these extreme traffic loads and still ensure service levels and optimal performance. And even if public clouds were designed to handle massive relational databases like those employed by Facebook, the bandwidth costs would be exorbitant and the latency would be unacceptable.

It appears the only option for traffic- and data-heavy social sites is to grin and bear the costs that come along with doing business. Hopefully, with some smart business people in the fold, they can figure out a way to make enough money that keeping the lights on represents only a small fraction of those costs.

Question of the week

Could a public cloud handle the performance and data needs of a site like Facebook?