Who will pay to fix the Internet's Interstates?
"Sherlock" Dignan had earlier fingered AT&T, which along with Verizon's MCI carries most of the Internet core's traffic. AT&T insists it's entirely a Googlefail.
Sources are telling our Sam Diaz that this has a lot of people rethinking their cloud strategies. SaaS planning will also take a hit.
Google planned for its growth carefully, and probably has more inter-city fiber capacity than any other IP service out there, more than some big ISPs.
Most of us are not so independent of the core. Google's problems are also proof that, even if you build independence into your business strategy, you can still run into problems.
Core Internet services, most delivered over fiber cable and optical switches, are the Interstate highway of today's Internet. Thanks to what I call Moore's Law of Fiber (and technical types call wavelength division multiplexing), the cost to move these bits keeps declining.
But this transport remains an essential service, and with prices continually dropping this service is in fewer-and-fewer hands. Profitability is also low, especially compared to licensed cellular, where service providers gain control of the content they transfer.
Right now core services are at the heart of their useful life. Costs of current deployments are bound to rise as systems age -- that's why the old bathtub curve tops this post.
We are heading for a crisis, and if this little outage wakes us up to that fact it's a very good thing.
Everything we do online -- downloading open source, running SaaS services, clouds, or simply reading this page -- depends on core services, which have become an essential utility.
Those services need to be redundant, they need to keep scaling, and they thus will need continuing investment. Just like the Interstate system.
Now there are two ways to fix the Interstates. We can share the cost and treat them as public utilities. Or we can sell them off as toll roads and pay for them as they're used, along with ample profit to the operator.
There is a lot of debate right now over whether the Interstates should become toll roads, but it's important to note that core Internet services already are. That model generally works well, today.
But for how long?