Can Linux Put PalmSource Back On Top?

Last December, PalmSource announced it had acquired mobile Linux vendor China MobileSoft and planned to use the company's technology to migrate its Palm OS to a Linux kernel. In the months following the announcement, Linux has enjoyed a burst of momentum, serving as the engine for a growing number of next-generation mobile devices. Motorola is well on its way to delivering on its promise to roll out 8-10 Linux-based phones this year, TrollTech says it is now working with 20 device vendors making Linux phones, and PalmSource in July landed its first major new licensee in years, electronics giant LG, with an agreement that looks unmistakably like a Linux play.

It&'s too early to tell how the leap to an open-source kernel plays out for PalmSource; for now the company is still fighting an uphill battle against declining sales of traditional PDAs and a series of delays that have plagued handsets running Palm OS Cobalt, the latest version of its mobile operating system. Today, even though PalmSource has refocused its energies almost entirely upon moving the Palm OS to Linux, the first devices running on its new platform are probably two years away.

Yet as PalmSource executives begin to speak more freely about the company's reasons for adopting Linux, it is becoming clear that an open-souce future could be very good to the Palm OS -- while PalmSource itself could play the same pivotal role in the Linux mobile device market that Red Hat played during the early years of the Linux server market.

Already A Success, But …

Many people find it surprising that worldwide Linux smartphone sales (as Gartner defines the term) already outsell those running Windows Mobile for Smartphones, Palm OS, and BlackBerry OS combind . Other ways of carving up the notoriously slippery “converged device” pie show Linux and Microsoft more evenly matched, however, and the fact remains that Symbian OS still dominates them all, claiming more than two thirds of the smartphone market.

Even so, the spike this year in the number of Linux-based 3G handsets has shifted some heavyweight mobile software companies in to full-hyperbole mode, such as OpenWave's comparison of mobile Linux to DOS during the early years of the personal computer market. Setting aside the hype (which is a familiar and unpleasant sight to desktop-Linux partisans), what are the fundamental forces transforming the world's most popular Web server OS into a killer mobile device platform? Do those forces justify PalmSource betting the farm on a risky strategic shift after years of work on Palm OS Cobalt, the proprietary Palm OS release created to compete against Windows Mobile and Symbian OS?

Simply put: Will Linux succeed as a mobile device platform?

The Kernel Counts

The best way to answer these questions, and to gauge the impact a Linux-based Palm OS will have on the mobile devices you use within the next few years, is to look at the convergence of mobile phones and personal computers as a story with three chapters.

The first chapter opens in 1999: Palm, Handspring and Kyocera were inventing the devices that created the “smartphone” category. These phones shared the proprietary real-time operating systems (RTOS) of their successful Palm PDA siblings, which were admirably efficient in their use of scarce memory, processor, and battery resources. But as providers built out high-speed wireless networks and microprocessors continued to obey Moore's Law, a "smartphone" meant more than just an organizer with phone and messaging capabilities. Multimedia and multiple wireless networking demands marked the beginning of the second chapter; the story now focused on a few favored OS architectures, built to deliver full, pre-emptive multitasking -- not unlike the desktop computer operating systems of the mid-90s. Despite a great user interface and the resounding commercial success of the Treo smartphones, PalmSource was starting to fall behind in the feature race against Symbian and Microsoft. Clearly, the Palm platform's RTOS roots could not carry it into the 3G era.

Enter the Palm OS "Cobalt" release. At the time PalmSource introduced Cobalt in early 2004, the company had every reason to believe the platform could keep up with its nearest competitor, Windows Mobile. Palm OS Cobalt 1.0 was a modern and elegant operating system that didn't carry generations of baggage from legacy chip architectures or vestigial desktop operating systems. But a third chapter in the smartphone story had already begun, creating first a crisis and then an opportunity for PalmSource.

Product Marketing Director John Cook is frank about the obstacles PalmSource and its licensees have faced getting Palm OS Cobalt handsets to market. Although Cook said some licensees' product-timing decisions proved very frustrating for PalmSource, he said it also became clear that "as a smaller company we were managing many, many moving parts."

The company's struggle to provide reliable hardware drivers typified the problem PalmSource faced. With Cobalt's new middleware framework running on a proprietary kernel that had never seen silicon outside of a lab, either PalmSource or its licensees had to create completely new drivers for every smartphone hardware element.

"Drivers became a critical time-to-market issue," said Cook. PalmSource simply lacked the resources to develop a monolithic, proprietary platform that could compete against similar offerings from Symbian and Microsoft. "The realization for us," Cook said, "was that we needed to find a leverage point." PalmSource also wanted to give more flexibility and freedom ot its licensees as a differentiation point from its brawnier competitors.

An "Open" Revolt

Fast forward to the present day. PalmSource clearly found its "leverage point": Linux. In Asia, which is both the largest and most advanced mobile phone market, handset makers are reluctant to pin their futures on proprietary platforms. That reluctance seems to be spreading: Except for Nokia and Sony Ericsson, both of which are committed to Symbian, every major device vendor now plans to develop phones on a Linux-based platform.

These vendors cite a variety of advantages to adopting an open source platform over a proprietary one. Interestingly, the lack of licensing royalties appears to be the least important advantage to most of them. Motorola, for example, which continues to ship Windows and Symbian phones but divested its ownership stake in Symbian two years ago, notes the advantage of choosing launch dates based on the company's own preferences rather than being tied to the OS developer's next major OS release.

Korean Linux phone pioneer E28 points to another benefit: the fact that closed, monolithic platforms often impose proprietary standards that restrict a company's freedom to differentiate itself from its competitors. Also, according to MontaVista Software&'s Peder Ulander, many of the company's mobile Linux platform customers appreciate the ability to unify their product line across a single platform.

Linux also provides the right technology mix to succeed in today's mobile market. Its open, modular architecture make it equally well-suited for the high-end smartphone segment as well as the much larger “feature phone” market, which includes devices that aren't quite as powerful as smartphones. For these devices, building a Linux-based OS can streamline the product development process while also meeting a company's interoperability requirements across its line. Even the fact that Linux powers so many mobile operators&' network infrastructures can lead to improved interoperability, while allowing companies to leverage experienced developers and proven standards. What will consumers see as Linux plays a far more visible and important role in the mobile market? Perhaps the most important benefit they will notice is a wider, more interesting array of phones to choose from. While Microsoft and Symbian often manage the complexity of their mobile platforms by limiting hardware support, the large Linux community can add new hardware support with astonishing speed -- often without requiring the phone makers to lift a finger.

What Can Palm OS Do For Linux?

While Palm aficionados hope that PalmSource&'s Linux bid will even the long odds the company faces as it competes against the likes of Microsoft and Symbian, it may be the Linux devotees who actually have the most to celebrate.

Cook and other PalmSource spokespeople are humble when describing the learning process they are going through as they engage with the communities surrounding Linux and other open-source software. Yet they are not shy about noting the contributions they expect Palm OS will make to the growth of Linux as a mobile device platform.

Despite increasingly positive market indicators. Cook said he believes that Linux is “still trying to get its legs.” And while Linux vendors like MontaVista and TrollTech mean that phone vendors don&'t have to roll their own system services, middleware and application stack, applications remain a weak point for mobile Linux -- much as they have been for the Linux desktop.

It's a small wonder, then, that MontaVista is enthusiastic about a new partnership with PalmSource. Despite the fact that MontaVista&'s top-to-bottom Mobilinux platform includes its own application stack, the prospect of MontaVista Linux-powered phones suddenly having access to more than 30,000 Palm OS applications is clearly a huge advantage.

The third chapter of this story is still being written, and the ending is far from certain. Given the advantages the Palm OS will enjoy, however, due to its new technical foundation, market dynamics, and the Palm platform&'s large and loyal following, Linux could very well turn out to be the lead character -- and this tale may very well have a happy ending.