Thus began my strange fascination with all-in-one, integrated kiosk platforms. Over the past few years, other hardware makers have come forward with compact solutions designed for self-service applications. But in my opinion (which is all you get here, after all), none matched the elegance and simplicity of the original X40. Even IBM's own successor, the X41, came up short. While the newer model had a rugged bezel and more power under the hood, a few physical flaws and the occasional unexplained stability problem kept me from recommending it to customers. With the supply of X40s practically nonexistent, any only a few half-hearted efforts on tap from other vendors building a turnkey platform, we had to turn back to the more traditional method of bolting a computer (or at least its innards) into the kiosk cabinet.
Now, I have nothing against this practice. It's what most of our customers have relied on for years, and there are many experienced kiosk vendors out there today who build high-quality kiosk systems using commodity computers and components. But from a technical support and product lifecycle perspective, this creates a few problems that could be avoided with an all-in-one solution. Among the concerns:
- Lack of hardware consistency - A Dell is a Dell is a Dell, right? Wrong. In fact, buy more than a few (supposedly identical) systems from Dell at once, and take them apart. More than likely you'll find that the motherboards, CPUs, RAM chips and hard disks are all slightly different from one another. While this isn't really a problem most of the time, I know there have been cases where a particular combination of parts was much more failure prone than others, or more difficult to troubleshoot for some reason. Lower-volume, all-in-one solutions necessarily contain more custom parts than mass-produced PCs, but every copy of a particular all-in-one device will be practically identical. Plus, vendors typically commit to offering the same model for 3 years or more, with only minor changes made along the way (e.g. faster CPUs).
- Not designed with self-service in mind - The thing that I like best about the all-in-one units (X40 aside, ironically) is that their manufacturers know that they'll be placed in hostile environments where they'll be kicked, punched, attacked with food and drink, and otherwise accosted. Of course, custom manufactured kiosks are designed to great tolerances as well, but too often they rely on a regular desktop PC inside, which most definitely was not built to these standards.
- Too much reliance on the integrator - Between the two points above, a kiosk integrator has to take many precautions and tie up a lot of loose ends to harden a system to the point where it can be deployed to a public environment. Again, there are companies who do this every day, and have deployed thousands of units this way. But they have to manually check every single device, ensuring that cables between monitor, touch screen and any other peripherals are correct and secure, and that the computer is mounted in a way that it can be shipped securely without getting damaged. This becomes especially challenging when the underlying components (e.g. LCD monitors and touchscreen controllers) need to be changed due to supply chain issues.
- Reliance on a single vendor for support and spare parts - Industry standard parts like CPU, RAM and hard disks are cheap and easy to come by, but if your all-in-one unit's motherboard is damaged by a power surge or physical trauma, you'll have to go back to your vendor as the sole supplier of spare parts. In addition to having only a single source of spare parts, you'll have to be content with the vendor's pricing policies, which may reflect their monopoly position -- especially in the latter part of the product lifecycle.
- End of Life - Relying on a single manufacturer also means that you have to trust their product lifecycle schedules. They may say they're going to produce the same hardware for 3, 5, or 7 years (which good vendors will stick to), but if they stop, you're not exactly left with a lot of options.
- Cost - All-in-one solutions are typically more expensive than a solution built with commodity parts. They contain custom hardware, they're produced in lower volumes, and they often use higher-end parts than commodity PCs, so of course they're going to cost more. While my experience has been that they make up for this with lower support costs, your mileage may vary.
It also looks like IBM isn't the only one noticing that there might be a growing market for this type of machine. Elo recently announced its latest entry in the all-in-one arena with the Elo 1529L TouchComputer. With a form factor and integrated peripheral options similar to the Anyplace, I expect that this will be a strong offering as well (though we haven't done extensive testing on it yet). Rumor has it that NCR will also be providing an Anyplace-alternative soon, which is good news since their current offerings (namely the EasyPoint and Personas) don't really play in the same space. And I wouldn't be surprised if POS manufacturers like Wincor-Nixdorf and Fujitsu get into the act with all-in-one kiosk offerings based on their existing hardware lines. (In my opinion, the current POS hardware from these vendors is a bit underpowered for multimedia-intensive kiosk applications, even though they usually run on standard Intel or AMD platforms.)
Is an all-in-one unit a good choice for your kiosk project? If you're shooting for a high-availability system that will be used for retail self-service and merchandising, then yes, it probably is. The most common peripherals (like the aforementioned bar code scanner and MSR) can be provided directly from the manufacturer, and other devices like thermal printers can be added just as easily as to any other kiosk computer. From a performance standpoint, you can't get an Anyplace (or any other all-in-one that I'm aware of) with the absolute fastest CPU, but that really doesn't matter for 99% of the self-service applications being developed today. And with a fixed hardware platform, if you test your application and find that it works properly on unit #1, you can be fairly certain that it's going to work properly with units #100 and #1000, even if they don't get deployed until two or three years from now.
Finally, after having deployed hundreds of kiosks from different vendors, I can confidently say that it is much less expensive to maintain all-in-one units over time. Between lower hardware failure rates and features that prevent many of the most common post-deployment problems (like touch screens needing recalibration, BIOS settings getting reset after power surges, and cables getting snagged or unplugged), I strongly believe that all things being equal, an all-in-one is going to have higher availability, better reliability, and less costly maintenance than a traditional kiosk during its lifetime. Of course, this doesn't mean that the services provided by traditional kiosk integrators are no longer needed. Their logistical expertise, service offerings and design capabilities still make them a valuable partner to anybody planning to deploy more than a few devices. Only instead of putting desktop PCs inside, perhaps they'll be bolting all-in-ones on top.