IBM’s acquisition of Red Hat got me thinking about how the market for commercially supported Linux distros is changing. IBM is trying to find a foothold in a maturing market dominated by AWS while the market for enterprise data centers is shrinking. So, where is Linux being used (or will be used), and what’s changing in those spaces?
To be clear: this is about commercial Linux distros, not upstack offerings like OpenStack, OpenShift, Kubernetes, etc.
Linux won the cloud, but clouds will increasingly develop their own distros
AWS Linux is the best example, but their work is just more publicly accessible than Google’s. Google has long used their own distro1 to power their internal cloud (and their famous network), and they’ve earned significant additional experience with consumer Linux on mobile devices and laptops (via Android and Chrome OS). As they grow their public cloud, what would stop them from introducing their own distro?
The holdout here is Azure. Microsoft long ago backed off from its attempts to push Windows everywhere and is instead pushing their applications and services. This appears to make them the least likely to develop their own Linux distro, though it’s possible they would feel forced to acquire one if Google makes a public move.
Linux distros will maintain a large and growing footprint in the cloud for some time to come, but that growth will likely decline relative to overall cloud spending growth.
Increasingly, the OS doesn’t matter in the cloud
Serverless is likely the future of the cloud. By definition, that eliminates customer/developer control over the OS, leaving cloud providers to use their OS distro of choice. Increasingly, that will be a cloud provider’s own distro.
Like serverless, cloud providers will host their Kubernetes-as-a-service offerings (as well as their DB and storage services) on their house-branded OS distro. Cloud customers have already shown greater interest in having their cloud providers do the maintenance and upgrades of their many Kubernetes clusters than in what specific distro those clusters run, and services like AWS’ Fargate further separate Kubernetes users from the OS and VM.
There’s still a question of what goes into the container, but priority in that space is minimalism (think Alpine Linux). Enterprises will spend significantly on support contracts in this space, but those will likely go to the cloud vendors, rather than the OS vendor.
Linux distros will maintain a large and growing footprint in the cloud for some time to come, but their five year growth outlook is likely underperform the overall cloud market.
The enterprise private cloud/on-prem/colocated market is changing and hotly contested
Representing the enterprise IT view, Gartner predicts:
Twenty-eight percent of spending within key enterprise IT markets will shift to the cloud by 2022, up from 19 percent in 2018, according to Gartner, Inc. Growth in enterprise IT spending on cloud-based offerings will be faster than growth in traditional, non-cloud IT offerings. Despite this growth, traditional offerings will still constitute 72 percent of the addressable revenue for enterprise IT markets in 2022, according to Gartner forecasts.
The drumbeat pushing enterprises to public cloud is continuing, and Gartner is joined by others predicting 21.7% growth rate for public cloud IaaS, which is about 65% faster growth than colo infra spending. Yes, there are analysts singing about continued growth in the on-prem and colo markets, but there can be no argument that public cloud offerings will eat a significant portion of spending here.
There are some shining examples like Dropbox that grew up in the cloud and are now racking their own servers, but these could become the industry’s white whale. In addition to being rare, companies that have the skill and stomach to move from the cloud to their own infrastructure—and then actually execute that move—also probably have the burning desire to build bespoke solutions (examples: network fabric and data center tooling), making it difficult for distro vendors to capture revenue from those moves.
Even if all new growth moves to the public cloud, this market represents a large business for Linux support contracts and it will remain so for several years as enterprises depreciate out their investments. Growth here will depend on companies seeing value in truly owning their own infrastructure, and with use-cases that offer clear cost advantages that can justify infrastructure build-outs that need to be amortized over five to ten years or more.
Linux won the handset market and is a fierce competitor in the consumer laptop space
But those wins are attributed to Android and Chrome OS specifically, not Linux generally. Dell’s Project Sputnik sales appear minuscule compared to their Windows and Chrome OS sales, using one vendor as an example2.
The investment necessary to gain a substantial foothold in these markets and compete against Google, Microsoft, and Apple is substantial (a reality Canonical conceded in early 2017). Non-Google Linux distros will still have a large footprint, but without substantial growth opportunity.
Clouds are moving to the edge
Lambda@Edge and CloudFlare Workers anticipate the emergence of IoT and 5G, and may be a great opportunity for cloud growth, but these serverless offerings will further diminish the visibility of the OS to developers.
Growth opportunities for Linux distros here will be due to an expected proliferation of new client/IoT devices that might run Linux.
Everybody is talking about IoT and 5G, but nobody knows what it looks like
This market has been anticipated for years and is already being strongly contested by a number of entrants. Google’s Brillo and Weave (now Android Things and Nest Weave) show the company’s intentions to own this space, but Ubuntu Core and many others have an opportunity here as well.
Android’s broad adoption may give Android Things a leg up in this space, but many are growing weary of Google. That might yield more opportunity for other Linux efforts like Ubuntu Core and Mozilla Project Things. Additionally, iOS users traditionally spend more than Android users do, which could further incentivize device makers to look at alternatives to Android Things.
Technically, they use their own kernel, previously in RedHat, now in Debian. I’m leaping from there to say they “use their own distro” because the stack of software running on top of those, including Let Me Container That For You is vertically integrated and nearly wholly owned by Google. ↩︎
Additional data points come from Google, which was a public advocate of desktop Linux. Google’s version of Ubuntu, Goobuntu, was said to have been used nearly half the company’s 20,000 employees in 2009 (the company switched to a Debian-derived distro in 2017-2018). By 2013, Macs were used by about 43,000 employees (that’s almost all employees at the time), and new employees are apparently being pushed to Chromebooks. ↩︎