One of Linux’s greatest and most distinguishing traits is its adaptability. Compare how it started and how it’s going, and you’ll see what I mean.
Linux began its life as a humble experiment in Unix porting, but from there it quickly became a popular kernel for server OSes due to its low cost (“free” is pretty low) and customizability. From there it morphed to power network appliances (think routers) and Android, proving it was lean enough for a whole gamut of embedded systems.
It’s easy to forget that Linux’s proliferation to every class of computing device has not concluded its evolution. With that in mind, I want to take a moment to not only catch up with its changes, but to potentially get ahead of them.
Don’t take this as gospel truth, though. This is just me expanding on an intuition I had while studying technology trends. Plus, the beginning of the year has me, like many people, in a prognosticatory mood.
The Penguin and The Man
As a Linux desktop user, I’m not always consciously aware of this, but it’s true nonetheless: Large tech companies are the strongest hand on the helm of the good ship Linux. It’s no accident that some of the biggest tech players — including Microsoft, Google, and Amazon — are among the biggest Linux Foundation donors.
This makes sense, since some of these companies’ most profitable products and services are completely dependent on Linux to operate. For example, what would Android be without Linux?
But that dependent status works both ways — because the Linux developers don’t charge for their hard work, they need donation dollars to pay the bills after banging out kernel code all day. There’s an implicit incentive, then, for the Linux kernel devs to prioritize addressing the interests of their donors.
Slowly, this skews the Linux kernel in the direction that business needs dictate. This isn’t a value judgment, just a characterization of reality; and in the grand scheme of things, it’s certainly a preferable reality to one without money for Linux development.
Linux Is Many Things, IoT Will Be Most of Them
A few days ago, it hit me that the future of Linux is the Internet of Things (IoT). To briefly define this category of devices, IoT is a class of embedded headless devices designed to be Internet-connected appliances. Consumer tech fans will best know IoT as home gizmos like smart assistant devices (e.g. Amazon Echo), home security cameras, and the like.
However, IoT devices are significantly more prevalent in the private sector. When a single new manufacturing plant comes online, for instance, its owner might install thousands of network-connected sensors. These then blast all their data back to company HQ for big data analysis.
While there are several reasons why I foresee IoT as the primary sustainer of Linux development, the main one is that IoT is the most varied device type today. Sure, there are lots of network appliances and servers out there. But with the plethora of IoT products available today, there’s no contest anymore. As new networked gadgets emerge constantly, IoT shows no sign of slowing down. In short, this means there will be a constant stream of new kernel modules that need writing.
There’s more to IoT’s looming influence over Linux development than the exploding number of new devices, though. Just as importantly, IoT is the most sorely in need of more secure development. The security of IoT devices on the whole has been abysmal for years, as I have covered in the past.
To recap, what tends to cause problems is that, beyond users seldom (if ever) logging in to them to check logs and software updates, IoT devices are often made by companies with little software development experience. Manufacturers of traditional analog products that are now sprinting toward the IoT gold rush haven’t learned the hard lessons of writing an operating system that the veteran developers have. Quite simply, IoT will garner most Linux developers’ attention because it will demand the most work to get on par with Android phones, Linux desktops, and Linux servers in terms of security.
But it’s not just that IoT’s star is rising. The Linux constellation’s other stars are setting. This, I think, will be most dramatic in Linux’s historically competitive server showing. Born as a pet project, Linux had a modest following until it caught the eye of for-profit companies because of its capable server computing. To this day, an enormous percentage of all servers run Linux.
To be more specific, it’s not that Linux is going anywhere on the server, but it may soon be beyond recognition. The new trend of “serverless” computing, if it continues its sharp ascent, may send Linux’s bare-metal server deployment share into decline. As cloud migration starts to slow, cloud service vendors like Amazon Web Services must compete more aggressively with rivals. An emerging tactic for doing so is by offering ever more convenient deployment options, and serverless computing is among the latest of these.
Server Distro Consolidation
In a serverless model, the cloud vendor abstracts server deployment and maintenance away from the customer, leaving only the application and database configuration for them to worry about. Why would you risk your server misconfiguration throwing a wrench into your plan when you could run disembodied application code?
Of course, serverless doesn’t really make the server disappear. Rather, the Web app creator just doesn’t have to deal with it. The cloud vendor is still installing an OS for that app to run on, and for sheer customizability and cost savings it probably is Linux. Actually, it’s probably doubly Linux: Linux on the raw hardware to run the hypervisor, and Linux in each of its virtual machines. But if enough Web startups opt for serverless setups, and the cloud market stays concentrated in few enough hands, most server Linux will be chosen, configured, and deployed by a few cloud providers.
This dramatically scales back Linux’s server diversity. Less companies installing servers means less Linux server distros will thrive. Cloud vendors will pick the handful they like and leave the rest. The end result is fewer server distros with high install bases.
So where does that leave Linux’s other holdings? The Linux desktop, as vibrant of a community as it engenders, is still niche among desktop users. Computer user population growth may help the Linux desktop accumulate devotees in absolute terms, but as a percentage, we’re roughly where we’ve always been. Where exactly I think the Linux desktop is going is a topic for another day. For now, all I will say is that no evidence I’ve seen suggests that “massively up in popularity” is it.
Embedded devices that aren’t IoT don’t look promising for Linux either. True, routers will always need to run something. But there’s a distinct possibility that Linux’s days as the leading mobile player are numbered. Every now and then, Google’s Fuchsia project surfaces to publicize a new milestone before quietly receding below the tech news surface.
Following the Penguin Tracks
It bears noting that, with as erratic and unforeseen a journey as Linux has had to this point, many a pontification has surely been wrong before. Will the dynamism of open source continue to propel it to new heights, or has it lost steam and gotten lost in the cloud?
Either way, I’m curious to see where our penguin friend wanders to next.