Google as CDN
Craig Labovitz, at Arbor Networks, a company that makes packet inspection gear, released a great article about how much traffic Google serves.
It has a few key points that the content delivery, content, and telco companies should note:
- Google is deploying cache servers in ISPs
- Google generates upwards of 5% of all internet traffic
- Upwards of 60% of Google’s traffic is delivered via peering
None of this should come to any surprise to anyone who has talked with me about how the Internet works, or how CDN
works. Anyone who generates a lot of traffic is a good candidate to take advantage of peering with end-user ISPs. Doing so will help improve quality and reduce cost to the ISP. Google, by definition, is a CDN
since they are routing traffic and server usage to a best server based on a user request and taking into account many factors in this decision.
But why is this controversial? Peering by large content owners and CDNs
has been happening since the dawn of the internet. CDNs
have been building and growing for over a decade.
The lesson here, if anything, is that we are still at the very beginning of the true revolution of internet usage, since we still get so little content, in such short bursts, at such low resolution compared to what could be possible in the future. Web pages are still small, the percentage of total video consumed at HD (1080p) resolution via the internet is still small. Interactivity and applications are still primitive compared to the immersive potential of the future – we still can’t watch a live broadcast of U2 in 1080p flawless quality from a concert in Europe, or have our doctors do a check up via videoconferencing so we don’t have to trek through a bad weather day to a doctor’s office.
Google’s efforts, while mildly interesting, are still really only the beginning. Good job to them for making large scale work to their, and our, advantage.