August 27, 2004

Survey: Linux Has Long Way to Go in Retail

By Evan Schuman, eWEEK

Linux penetration into retail isn't nearly as significant as commonly believed, according to a new study from industry analyst firm Venture Development Corp. (VDC).

"There's a lot of talk, a lot of news around Linux and a lot of hype," said Mike Liard, senior analyst at VDC and a co-author of the report. But its survey of retail IT execs found Linux in barely 2 percent of the POS (point of sale) systems, compared with 43 percent for Windows, 19 percent for various proprietary/in-house customized operating systems, 17 percent for IBM (primarily AS/400), 9 percent for various Unix flavors and 10 percent that didn't answer.

"I had expected to find that Linux had about five percent of the market, but this is on par with what's really out there," Liard said. "Linux is struggling in this space."

VDC analyst Taylor Smith, another report co-author, said he was also taken aback by Linux's weak showing. "I was a little surprised; I was expecting Linux to have gotten a little more traction than it got," he said. "You will hear a lot of talk about Linux, but it still appears that Windows will remain the big man in charge for the next five years."

Linux does still have a shot to ultimately generate a more substantial installed based within retail, Smith said. There are several reasons why a lot of retail POS systems are going to be upgraded in the next year or two—including compliance labeling and a move to have U.S. retailers move from the current 12-digit UPC to 13-digits—and those upgrades make it easier for a retailer to evaluate new operating systems.

In other conclusions about retail technology in the report, VDC found that bar-code label generation software sales hit $76 million last year, with an 11 percent annual growth rate projected through 2007. It also found that the breakdown between off-the-shelf bar-code software and customized in-house versions remains fairly evenly split (53 percent were off-the-shelf and 47 percent custom).

The biggest bar-code issue will likely be data synchronization, Smith said. In testing, retailers still find that much of their data synchronization efforts—some say as high as 60 percent—generate some sort of error. This is only going to be a bigger issue as retailers increasingly move to self-checkout systems, which add one more database for data to be synchronized from.

Another point of data-synchronization difficulty will be with RFID. Although widespread product-level RFID identification is still several years away, RFID is still going to be prevalent at the pallet and crate level as well as isolated tests of RFID at the item level. Liard said that means that the challenge of data synchronization will not be how to transition from bar code to RFID, but how to increasingly co-exist with the two for many years.

Despite industry events such as Wal-Mart's 2005 deadline for its top suppliers to move to RFID, Liard said he doesn't to see any full-scale RFID rollouts next year.

Despite the age of bar-code technology, it's neither as saturated nor as advanced as it's going to get. A major North American UPC code effort called 2005 Sunrise is an attempt to move bar-code systems from 12 digits to 13 digits. European retailers, for example, are often at 14 digits because they're so frequently used in other countries, particularly within Europe.

"With the amount of hype and market information out there, end-users were mislead early on about RFID," Liard said, adding that RFID was always intended to complement—not replace—the bar code. "We're not close to the highest point of bar-code implementation yet," he said.

One of the other problems that cropped up in the VDC research was the fickle nature of RFID installations. "RFID still has some physical limitations. Each install is inherently unique because no two warehouses are exactly alike," he said. Read accuracy is an issue, along with adjusting antenna arrays and tweaking tag configurations, he said.