Tapping the RFID Data Flood

There's concern that RFID will lead to a flood of data that will overwhelm IT systems. Can you explain the importance of RFID data management?

RFID represents a fundamental shift in the ease, ability and speed at which you can gather information about events in any operation or supply chain. Conventional systems, such as bar code printers and readers, were designed for human processing that limited transaction volumes. RFID can automate these processes and has the potential to generate orders of magnitude more information.

The most cited example has been the estimate that Wal-Mart's in-store implementation will generate about 7.5 terabytes of RFID data a day. Certainly we haven't seen this scale of implementation yet, but the example speaks to the possibility that the volume of information could be overwhelming in larger operations. Therefore, it's important to have new data management approaches.

Not all RFID events are relevant to ERP and BI systems. How do you filter the information that's relevant for business use?

This is one of the big challenges of RFID systems. Probably 99 percent of RFID data may not be relevant to enterprise applications, but it's critical to process the intelligence within that data. The fact that a case or pallet just moved past RFID scanner number 19 is not important. However, if you can somehow derive from that data that the case or pallet actually wound up going onto the wrong truck, that is valuable.

This information can't easily be gained through conventional systems. So the filtering of that data is where new developments in complex event processing come in. The technology will become more prevalent as enterprises adopt RFID and need to figure out how to discover patterns and meaningful information among the huge volumes of RFID data.

Is the software industry keeping up in supporting such requirements?

Yes, it is, but I don't think a lot of people are aware of how advanced the technology is today. This technology was actually developed for other industries. For example, one of the hottest areas on Wall Street is algorithmic trading — automated trading supported by complex event processing. This was built for the largest trading floors in the financial services industry, and it's the very same technology required by RFID systems to process high-volume event streams in real time.

When will the use of RFID become the norm?

I think RFID will become dominant within five years, but I see it as an evolution because there are so many things that change with RFID. For example, the physical construction of buildings as well as how you train people often need to be altered for RFID because you can automate more of the environment. These are things that don't change overnight.

RFID can also fundamentally change the ability to implement the intelligent enterprise. It starts to get interesting as more predictive applications enabled by RFID and BI show up. Of course, it's going to take a while to figure out exactly what all those really interesting new business models and business processes are, but you can be sure they'll be more real time and more accurate.

OUTTAKES

What do you collect? Old running shoes — I'm a marathoner.

What's your favorite thing to do with your kids? Chase them at the playground, at the beach and in the yard. Once I catch them, I take pictures of them.

What college team do you root for? UConn, my alma mater and men's basketball champs in 1999 and 2004.