Scale Matters

By Kevin Ashton

As RFID moves from pilot to deployment, the thing that will most starkly separate success from failure will be the ability of systems to scale.

  • TAGS

Back in 2002, a U.S. company that was an early adopter of RFID technology was working on a lab pilot in which the reader kept crashing. It seemed like no big deal—the RFID team just turned it off and on again. But in deployment, this problem took center stage. The RFID team had to phone workers thousands of miles away and beg them to go and turn the readers off and on. Sometimes they did, and sometimes they didn’t. The deployment quickly ground to a halt.

Scalability—the ability to grow without causing ever-increasing problems—is rarely mentioned in RFID discussions. But any RFID pilot that won’t scale is doomed.

Mathematicians have a way to identify solutions that don’t scale: If the time it takes to solve a problem grows out of all proportion to its size, the approach is condemned as “brute force” and inelegant. RFID users should take the same view. Brute force may work in an RFID pilot, where a system can be carefully maintained by a few trained and motivated engineers working full time to keep things running, but it will not work in deployment. If adding twice as many RFID-enabled locations gives you four times as much trouble, your system does not scale.

The growth of enterprise-wide computer systems has taught CIOs most of what they need to know about scale. What matters now is to apply those lessons to RFID.

In some cases, this is easy. Everything from the RFID reader back is a computer system like any other. The techniques and technologies that work on other IT systems will work for your RFID system. Security, for example, should be as high on an RFID network as it is everywhere else, and industry-standard approaches such as Secure Sockets Layer and Secure Shell should be a requirement at deployment.

The same thing is true of basic networking. Network systems have grown, in part, because of smart technologies that allow automatic distribution of network addresses, remote maintenance and logging, and the ability to upgrade to new software automatically. All of these things can be applied to RFID during deployment.

But the front end of an RFID system—the radio interface between tag and reader—is a little more challenging, and new solutions must be found. Wireless networks are newer than other computer systems, and some of their problems, such as signal interference, are still being solved. But these issues are not completely novel. Lessons from scaling wireless access points can be applied to RFID. Processes for identifying and eliminating interference, such as site surveys, will work for RFID systems, too, and it will become important to detect and control unauthorized wireless devices.

There are, of course, a few areas where nothing is familiar—the intricacies of dense-reader environments, for example—and here it will be important to cultivate in-house or independent experts who can develop solutions that scale. But even in these instances, there is one familiar lesson that every CIO can—and must—reapply. In RFID, as in any other IT system, size—or, more precisely, scale—matters.

Kevin Ashton was cofounder and executive director of the Auto-ID Center. He is the author of a soon-to-be published book about RFID.