They're everywhere, but you rarely notice them: the millions of handheld devices -- often equipped with scanners -- that delivery people, store clerks, and hospital staff responders often carry to manage inventory, process orders, and verify delivery.
Nearly all of these supply chain-oriented devices run a version of the Windows Embedded operating system, which has had many names over the last decade. But within five years, the companies using these devices will have ditched Windows and moved to Android in one of the biggest industry platform shifts ever.
Before Windows Embedded came to dominate these specialty handheld devices in the mid-2000s, they ran a variety of operating systems. The best-known was from Symbol Technology, which been bought and sold several times over the years, and whose remains form the core of a company called Zebra Technologies.
Why the industry is looking to ditch Windows Embedded
The industry coalesced around Windows Embedded CE version 6, which debuted in 2006 and represents 99 percent of all embedded devices' operating system, says Mark Kirstein, senior director of enterprise software at Zebra, the company that makes most of those specialty devices.
But the industry largely skipped 2011's Windows Embedded 7 and almost totally skipped 2012's Windows Embedded 8, Kirstein says. He expects the same to be true for the new Windows IoT, a variant of Windows 10.
Version 8 and later aren't compatible with Version 7 and earlier, so developers have to create their devices' apps from scratch, he notes. Plus, Windows Embedded 8 didn't support the wide range of screen sizes and form factors that specialty devices need.
Kirstein says Windows Embedded 8 was merely a version of Windows Phone 8 minus the Xbox functionality and not designed for specialty devices such as for supply chain use. "It doesn’t have the enterprise management capabilities or staging/provisioning services -- and therefore was never adopted. It did not meet the requirements of Kohl’s, Walmart, Target, Home Depot, and so on," he says.
Microsoft is ending extended support in 2017 for Windows Embedded 6 -- the one that powers 99 percent of current devices -- and in 2020 for Windows Embedded 7. Thus, developers and device makers need to rewrite their apps for a new OS -- whether from Microsoft or someone else -- before then to be safe. Many devices are used in regulated environments where support is required.
The transition won't be cheap or fast: "It took Home Depot four years to migrate 28 apps," Kirstein recalls. Enterprises like Home Depot, Carrefour, Whirlpool, Jepco, Tropicana Resort and Casino, and the U.S. Air Force typically have dozens of specialty-device apps, and some of those apps hit 1 million lines of code. The transition can cost as much as $10 million, he says. With that level of investment, enterprises need to get really comfortable with the platform they adopt.
The shift from a focus on enterprises to consumers in Windows Embedded 8, and a similar focus in the Windows 10-based Windows IoT, has prompted the industry to look elsewhere. As Kirstein recalls:
Home Depot was the largest planned deployment of WE8HH [Windows Embedded 8 for Handhelds] , but never happened. I was driving/owning the Microsoft/MSI relationship during this announcement. [MSI is Motorola Solutions, which had bought Symbol and then got bought by Zebra later.]
The end result is Microsoft never delivered the OS on time when they committed to it, nor did it include the key requirements that were expected or needed for enterprise devices. Chipset vendors stopped supporting it, and then we were told we had to wait for Windows 10 to come out. Home Depot moved on, as well as dozens of other large enterprises.
Where did they move? "They are all rolling out Android now," Kirstein says.
Why Android attracts embedded device developers
More precisely, most are moving to the Android Open Source Project (AOSP), the part of Android that does not include the Google Mobility Services (GMS) component set. GMS is where Google makes money from Android by using apps like Gmail, Chrome, Google Drive, Google Now, and Hangouts to gather user data, then sells to advertisers and others.
Using AOSP provides developers several advantages, Kirstein notes.
One is the large base of Android developers to draw from. The apps that run on these specialty devices come from companies like Zebra, supply-chain software development firms, and the enterprises themselves that use the apps. Few developers know how to write for Windows Phone 8 or Windows 10 -- Microsoft's latest offerings -- but many know how to write for Android.
Another advantage is that developers can modify the Android OS to add capabilities and restrictions not supported in the core OS. For example, a developer can modify AOSP to set the time zone or enable Bluetooth connections without user interaction, to reduce the human errors that could mess up workflow or data consistency in logistics operations across a region.
But developers can't remove capabilities, so any Android app will run on these AOSP devices without modification. Why does that matter for a specialty device? Kristen uses the example of Best Buy, which has special product-scanning and inventory-checking handsets for employees. Those handsets can also run the Best Buy app and the Red Laser price-checking app, as well as access the Internet. Thus, employees can see what their customers see on their smartphones and use the same apps to comparison-shop.
That example made a lot of sense to me, given a recent experience I had at a Best Buy store where the app showed a lower price than the store's internal system. I let the clerk see the price on my iPhone to prove I was owed a discount; it would have been better if she could have done so herself -- then flagged the inventory management system -- on her own company device.
There's also some feeling in the industry that AOSP is a safer bet than Windows Embedded, Kirstein notes. He says that at its height, in 2007, Windows Embedded accounted for 0.01 percent of Microsoft's profit. Then-CEO "Steve Ballmer couldn't afford to pay attention to it," he jokes. That's why Microsoft let the enterprise focus end.
Plus, Microsoft no longer needs to use Windows Embedded as a "socket to the enterprise back end," Kirstein argues -- its decision to make its apps and services available to iOS and Android (often before Windows Phone) has let all devices feed into the core Office 365 and Azure services that now power Microsoft's revenues.
But Google makes no money on AOSP, so why is that a safer bet for enterprises' specialty devices? Kirstein believes Google needs AOSP to create a big enough ecosystem to funnel people into the GMS part of Android; he calls it a flanking strategy and notes Google uses it in noncore products to drive users to the data-oriented services where it does make lots of money. "Google needs the broad adoption to ensure enough people who pay for the whole thing."
Of course, Google isn't developing AOSP in any significant way; it has pivoted to GMS, where the money is. But the open source nature of AOSP helps it remain viable, Kirstein says: "Many large companies are so invested that they would carry the ball."
If GMS becomes required, the industry will adapt, Kirstein says -- as long as "Google can better address the corporate-liable approach." Google has taken some steps in that direction recently, such as through its Android for Work technology.
Beyond Android, Tizen might await
Of course, should AOSP wither away like Windows Embedded has, the industry will have to shift again. Kirstein sees only one possible candidate: Tizen, the open source mobile operating system backed mainly by Intel and Samsung.
Tizen has flopped in consumer devices, but Samsung is a huge player in specialty electronics and devices. If it adopted Tizen, that would create a huge base to attract developers and enterprises.
Of course, Samsung has been ineffective at garnering developer interest in its own platforms. It's a long-shot bet that would rely more on Google forcing the supply chain industry to find a new platform than on Tizen itself.
There's also no evidence that Samsung has an interest in Tizen for supply-chain-oriented devices. Although the company recently launched an Internet of things business (who hasn't?), it did not respond to InfoWorld's inquiry as to whether it saw Tizen as part of that effort. Translation: That's often how Samsung says no without saying no.