Magazine Article | June 1, 2003

Multifunctional, Thin Client POS?

Source: Innovative Retail Technologies

If you curse the lack of functionality at your POS, perhaps it's time to take another look at thin client solutions.

Integrated Solutions For Retailers, June 2003

The trouble with the entire hullabaloo about thin client architecture at the POS is the watermark it needs to achieve before retailers buy into it. Say what you must about that old die-hard IBM 4680 you're running, but admit it - they really had something there. For the first time, retailers could process sales with no programs or data at the POS station. If only it could be replicated today, on a modern platform that would actually let you adopt some of the cool applications retailers are capable of managing at the POS - things like Web and inventory access. POS software vendors are getting closer to giving retailers what they want, and with so much to gain by enabling thin client solutions, hardware manufacturers are in the game, too.

Java's Lean And Smart
At Wincor Nixdorf (Austin, TX), Director of Retail Solutions Development Doug Evans has a goal. He says he could get rich if his company could develop an unbreakable POS system that's thin as a palm pilot, less than $100, and that works offline. That draws a chuckle from the editor, but do realize that Evans says this with his tongue only half-planted in his cheek. Wincor is working closely with software providers tinkering with Java and other "smart client" platforms with hopes that when retailers like the solutions enough to dole out cash for them, they'll like them running on Wincor's Beetle hardware platform. In the meantime, Wincor isn't waiting for the grocery industry to ditch its IBMs and buy Beetles en masse. At Retail Systems this year, Wincor will hype its new communications hub, which allows IBM peripherals to run off of the Beetle. As long as grocery has had POS, IBM has owned it. Not a hardware manufacturer on the planet can snatch that away overnight, but as software migrates to open standards, it's a race to see which hardware vendor(s) will claim the lion's share of next-generation grocers.

"We think a 'smart client' deployment, which is very similar to what was done 25 years ago by IBM but using modern technology, is the answer," says Evans. He describes a small footprint kernel that can be deployed via flash or a spinning disk on the client, or loaded at boot time via the network. "A 'smart client' can be deployed in a flexible manner, meaning objects can be moved forward or backward between the client and the server, or the back office server to the enterprise servers, depending on the retailer's comfort zone and factors like network bandwidth," says Evans.

Barriers To Thin Client Adoption
Chuck Renfroe, professional services manager at NewBold, says software and WAN (wide area network) issues have hindered the development of thin client architectures, but those problems are being addressed by today's thin client solutions. "Whether or not it's buildable hasn't been relevant. It doesn't matter unless there's software that will make it usable. That's now beginning to take place," he says. The WAN, the means of communication employed by the solution, is also becoming less a problem. He explains that in some large organizations, such as hospitals, thin client has been commonplace for some time. But in this environment, everything is in the same facility-the server, the network, and all the terminals. In retail you may have stores spread across the country, but you need to give thin client stations access to the corporate host. "The limiting factor has been the means of communication, or the WAN," says Renfroe. "There are still places in which dial-up is the only means of connectivity, but that's becoming less a factor. The combination of a good network infrastructure and software availability is making thin client a reality."

As you might gather, hardware vendors know the primary driver of POS sales is the software. But hardware facilitates the return, especially with thin client applications, which are hardware dependent. "If you can lower and maintain the overhead cost of running the system, that will increase your rate of return, but that's not the primary reason you would deploy thin client," says Renfroe.

When it comes to Java's role in enabling thin client POS flexibility, Evans says he's seen some early applications that, in his opinion, aren't architected for deployability across the enterprise, but that it's coming around. "If you look at the top two or three software providers in the Java space, they've gotten there," he says. Cross-platform deployability is another benefit of Java. "Between the client, back office, and host enterprise, you might have three different operating systems. That doesn't matter to Java, it can be deployed across all of those," says Evans. "If there is a problem with Java, it's that it isn't very thin in terms of deployment on the client side. You can thin down the business logic quite a bit, but the bar is still pretty high to get the Java VM (Virtual Manager) GUI (graphic user interface) and all the infrastructure to support the JBM up and running on the client."

What's In It For You?
Payback on a POS upgrade is hard to compute. But vendors across the board stress that, by and large, retailers fail to realize the amount of money they spend keeping legacy systems running. When you analyze the ROI for deploying new POS, does what you spend maintaining legacy systems manage to get into the calculation? Are you factoring in the cost of maintaining hard codes, or changing them every time the business requires it? Evans estimates that in a typical legacy system, a hard code change-a tax adjustment, for instance-might require a 14-week turnaround. That sort of lag time hinders the retailer's ability to react in an agile manner and gain an edge on the competition when the competition is running better systems. If the retailer says it now needs to charge tax on a product and a parameter needs to be changed, it could be done in as little as 24 hours. If you must change code in a legacy system, even in an emergency, you could be looking at a 3 to 4 week turnaround.

No doubt the thin client approach is working for some retailers. You've read about some of them right here in this magazine. But how long will it be before today's applications are available on a system like the venerable 4680? As Evans says, "We're all trying to find the solution that fits the retailer's budget, satisfies their requirements, fits on the cash wrap, etc. Nobody's nailed it yet."