As one of the leading manufacturers of the equipment that routes data around the Internet, Cisco Systems is in good position to know just how many 0s and 1s go zipping around all day, every day. Today it released an annual analysis of how much Internet usage is growing on mobile devices, and the report produced some staggering numbers.

For example, Cisco estimates that the amount of data that was ferried to and from mobile devices last year was eight times greater than the data on all of the Internet in 2000. Global mobile data traffic is expected to see an 18-fold increase between 2011 and 2016. Not surprisingly, video is a big reason: Cisco expects there to be 7.6 exabytes of data flowing to mobile devices every month in 2016, about 70 percent of the total of 10.8 exabytes of data per month. (An exabyte is more than 1 billion gigabytes and equivalent to 250 million DVDs, if that helps you wrap your mind around it.)

But you might be surprised by the second-leading source of the expected surge in traffic. It won't come from people, but from machine-to-machine communications, or "M2M." Think of sensors in cars and in appliances, surveillance cameras, smart electric meters, and devices still to come, monitoring the world and reporting to each other and to centralized computers what they're detecting. The chart below, reprinted from the Cisco report, shows just how extreme the jump in machine-to-machine communications could be. It is expected to grow, on average, 86 percent a year, and by 2016 it is expected to reach 508 petabytes a month, or half a billion gigabytes.

Source: Cisco's Visual Networking Index Global Mobile Data Traffic Forecast Update, 2011-2016

This week's Gmail outage highlights the key problem with cloud computing: it means handing both your data and your infrastructure over to someone else.

Understandably, many businesses prefer to have more control. Struggling to repair an email server is, after all, marginally preferable to refreshing the Google maintenance page with both fingers crossed. So as Google tries to encourage more corporate customers to use Gmail, Google Docs, etc, expect the company to push the option to mirror data locally (something that's already possible through Google Gears). Microsoft has already announced that customers using the online version of Office 2010 will be able to store data in their own data centers if they choose.

Perhaps the retreat from the cloud dependence could go further still. Why not let companies switch back to using local servers whenever the main service goes down, as inevitably it will, from time to time. This might be technically difficult, but it doesn't seem impossible.

And, if your heart sank when Gmail fell it may be time to consider the benefits of an old-fashioned local mail server. Sure, you could spend hours configuring it or troubleshooting problems, but at least you'd know someone was working on the problem. And it would let you safeguard your own data and protect your privacy.

Danny O'Brien of the Electronic Frontier Foundation calls this "Living on the Edge" and argues that it may the best way to preserve certain freedoms in an age when more and more information is floating off into the cloud. (I saw him give an interesting talk on the subject at OpenTech 2008--you can see a rather shaky video of it here). For more on the possible dangers of cloud computing, also check out Cory Doctorow's latest column for The Guardian.

Most people leave a trail when surfing the Web. Information such as a computer's IP address can be traced back to users, or used to reconstruct a profile of browsing habits. Search engines amass large quantities of data on individuals. Though they don't store this along with usernames, researchers have previously shown that individuals can still be identified using this data.

People who want to avoid leaving this trail can turn to services such as Tor, an open-source system designed to muddy the path a user's data travels over the Internet (see "Dissent Made Safer"). But Tor struggles with slow network performance, and the service might be overwhelmed if too many users adopted it without also contributing resources.

Last week, at the 9th annual Privacy Enhancing Technologies Symposium, researchers described some more robust protections. They wondered if privacy protection could come from the ISPs responsible for the backbone of the Internet.

One project, anon.next, presented by Matthew Wright, who co-directs the iSec research lab at the University of Texas at Arlington, looks ahead to next-generation deployments of the Internet itself. In the event of a redesign of Internet architecture, Wright argues, proxies that help preserve anonymity could be built in. He envisions working with ISPs to determine points in the network where the proxies would be effective both in terms of protection and performance.

Other researchers are looking for solutions that could work on the Web as it is today. Barath Raghavan, a visiting assistant professor at Williams College in Massachusetts, along with researchers from the University of California, San Diego, and the University of Washington, suggest a protocol that could effectively hide a user's IP address within the rest of an Internet service provider's traffic. The researchers say that adding their system wouldn't hurt performance, and would work in conjunction with Tor and other privacy-protection services. They suggest that ISPs might be willing to add the protocol as a benefit to attract customers, similar to services offered by telephone companies that prevent users from being identified by called ID.

While ISPs are a logical place to turn for privacy help, events such as the passage of the Patriot Act in the United States, which made it possible for the authorities to demand information without a subpoena, make ISPs uncertain allies. The bottom line is, they're only likely to help if there's a large customer demand for privacy.

Most people think of online privacy as something most important for citizen journalists in countries with oppressive regimes. However, the number of business models that rely on the collection and sale of user data may for some people in this country to reconsider taking steps to protect it.