Google: Earth's third-largest ISP

The latest indication that Google is one of the most powerful players in telecoms: if it were an ISP – and if you measured ISPs by the interdomain traffic volumes – Google would be the third biggest ISP on the planet, and the fastest-growing.
 
That’s according to Arbor Networks, a network security/monitoring company that took traffic data from 110 ISPs worldwide and worked out that Google alone accounted for anywhere from 6% to 10% of global Internet traffic by June 2009.
 
A lot of that has been driven by YouTube traffic, and Arbor’s metrics don’t include internal provider services like IPTV or VPN or data from caches within provider data centers. But Arbor says that interdomain traffic is “a key metric for understanding Internet topology and the evolution of internet traffic patterns.”
 
More to the point, the Arbor data shows that Google is becoming increasingly self-reliant in moving that inter-network traffic.
 
In 2008, Google was using direct peering for less than a third of its global data traffic, with the rest being transported by IP transit service providers. As of last month, over 60% of its traffic uses direct peering.
 
Meanwhile, Google has been deploying its own Google Global Cache (GGC) servers within the same consumer networks that it peers directly with, which helps Google save on transit costs and improve traffic latency.
 
What does all that mean for the world’s “real” ISPs? Well, they won’t have to worry about Google getting into the proper ISP business, but Ryan Singel of Wired says the real Google threat comes down to Google’s overall plan to work with communities to build an open infrastructure that anyone can lease to sell internet services.
 
“If Google can come up with a partnering model that costs them little or uses a revolving fund [Singel writes], they could create a workable model for communities to get beyond reliance on companies like Verizon and Comcast. Using some of Google’s cash, a lot of it’s know-how, and citizen dissatisfaction with the current costs of not very fast broadband, municipalities could forge a viable alternative to the current system of begging and pleading for telecoms to lay fiber in their areas.
 
“Certainly, Google has the motivation and a pocket full of cash. It’s just a question of whether its innovation in transit can extend from its current efforts to rework how its packets get to consumers to remaking how networks are owned and built.”
 
And that’s particularly interesting considering Google’s planned fiber broadband access experiment and the US FCC’s current plan to make 100-Mbps broadband available on a nationwide basis. If the telcos and the government can’t make it happen in certain areas, perhaps Google will.
 
For all that, of course, the same dynamic is unlikely to happen in Asia-Pacific, if only because the regulators and governments that have already committed to national broadband plans have firm control over the process. Singapore’s IDA, for example, has already decided who gets to build the open infrastructure for its NBN project. It’s hard to imagine Google being allowed to come in and engineer a workaround with local partners.
 
Ditto Australia’s NBN project, although the current dithering and shouting between Telstra and the government over the details could conceivably delay rollouts enough to make some local players wish they could rope in Google or someone to help them put together a decent broadband service sooner.