Linux Still Rules IoT, Says Survey, with Raspbian Leading the Way

By Eric Brown

The Eclipse Foundation’s Eclipse IoT Working Group has released the results of its IoT Developer Survey 2018, which surveyed 502 Eclipse developers between January and March 2018. While the sample size is fairly low—LinuxGizmos’  own 2017 Hacker Board survey had 1,705 respondents—and although the IoT technologies covered here extend beyond embedded tech into the cloud, the results sync up pretty well with 2017 surveys of embedded developers from VDC Research and AspenCore (EETimes/Embedded). In short, Linux rules in Internet of Things development, but FreeRTOS is coming on fast. In addition, Amazon Web Services (AWS) is the leading cloud service for IoT.

 Eclipse IoT Developer Survey 2018 results for OS usage (top) and yearly variations for non-Linux platforms (bottom)
(Source: Eclipse Foundation)
(click images to enlarge)

When asked what operating systems were used for IoT, a total of 71.8% of the Eclipse survey respondents listed Linux, including Android and Android Things (see farther below). The next highest total was for Windows at 23%, a slight decrease from last year.

The open source, MCU-focused FreeRTOS advanced to 20%. Last December, the FreeRTOS project received major backing from Amazon. In fact, the Eclipse Foundation calls it an “acquisition.” This is never an entirely correct term when referring to a truly open source project such as FreeRTOS, but as with Samsung’s stewardship of Tizen, it appears to be essentially true.

Amazon collaborated with FreeRTOS technical leaders in spinning a new Amazon FreeRTOS variant linked to AWS IoT and AWS Greengrass. The significance of Amazon’s stake in FreeRTOS was one of the reasons Microsoft launched its Linux-based Azure Sphere secure IoT SoC platform, according to a VDC Research analyst.

The growth of FreeRTOS and Linux has apparently reduced the number of developers who code IoT devices without a formal OS or who use bare metal implementations. The “No OS/Bare Metal category” was second place in 2017, but has dropped sharply to share third place with FreeRTOS at 20%.

Other mostly open source RTOSes that had seen increases in 2017, such as mBed, Contiki, TinyOS, and Riot OS, dropped in 2018, with Contiki seeing the biggest dive. All these platforms led the open source Zephyr, however, as well as proprietary RTOSes like Micrium PS. The Intel-backed Zephyr may have declined in part due to Intel killing its Zephyr-friendly Curie module.

Eclipse IoT results for OS usage for constrained devices (top)
and gateways (bottom)

(Source: Eclipse Foundation)
(click images to enlarge)

When the Eclipse Foundation asked what OS was used for constrained devices, Linux still led the way, but had only 38.7%, followed by No OS/Bare Metal at 19.6%, FreeRTOS at 19.3%, and Windows at 14.1%. The others remained in the same order, ranging from Mbed at 7.7% to Riot OS at 4.7% for the next four slots.

When developers were asked about OS usage for IoT gateways, Linux dominated at 64.1% followed by Windows at 14.9%. Not surprisingly, the RTOSes barely registered here, with FreeRTOS leading at 5% and the others running at 2.2% or lower.

Eclipse IoT survey results for most popular Linux distributions
(Source: Eclipse Foundation)
(click image to enlarge)

Raspbian was the most popular Linux distro at 43.3%, showing just how far the Raspberry Pi has come to dominate IoT. The Debian based Ubuntu and more IoT-oriented Ubuntu Core were close behind for a combined 40.2%, and homegrown Debian stacks were used by 30.9%.

Android (19.6%) and the IoT-focused Android Things (7.9%) combined for 27.5%. Surprisingly, the open source Red Hat based distro CentOS came in next at 15.6%. Although CentOS does appear on embedded devices, its cloud server/cloud focus suggests that like Ubuntu, some of the Eclipse score came from developers working in IoT cloud stacks as well as embedded.

Yocto Project, which is not a distribution, but rather a set of standardized tools and recipes for DIY Linux development, came next at 14.2%. The stripped-down, networking focused OpenWrt and its variants, including the forked LEDE OS, combined for 7.9%. The OpenWrt and LEDE OS projects reunited as OpenWrt in January of this year. A version 18, due later this year, will attempt to integrate those elements that have diverged.

AWS and Azure rise, Google Cloud falls

The remainder of the survey dealt primarily with IoT software. Amazon’s AWS, which is the cloud platform used by its AWS IoT data aggregation platform and the related, Linux-based AWS Greengrass gateway and edge platform, led IoT cloud platforms with 51.8%. This was a 21% increase over the 2017 survey. Microsoft Azure’s share increased by 17% to 31.2%, followed by a combined score for private and on-premises cloud providers of 19.4%.

The total that used Google Cloud dropped by 8% to 18.8%. This was followed by Kubernetes, IBM Bluemix, and OpenStack On Premises.

Other survey findings include the continuing popularity of Java and MQTT among Eclipse developers. Usage of open source software of all kinds is increasing — for example, 93% of respondents say they use open source data base software, led by MySQL. Security and data collection/analytics were the leading developer concerns for IoT while interoperability troubles seem to be decreasing.

There were only a few questions about hardware, which is not surprising considering that Eclipse developers are primarily software developers. Cortex M3/M4 chips led among MCU platforms. For gateways there was an inconclusive mix of Intel and various Arm Cortex-A platforms. Perhaps most telling: 24.9% did not know what platform their IoT software would run on.

They did, however, know their favorite IDE. It starts with an E.

Further information

More information on the Eclipse IoT Developer Survey may be found in this blog announcement by Benjamin Cabé, which links to a slides from the full survey.

This article originally appeared on on April 30.

Eclipse IoT Working Group |

STMicroelectonics Sensors Achieve Validation for Alibaba IoT OS

STMicroelectronics has announced the validation of its LSM6DSL 6-axis inertial sensor and LPS22HB pressure sensor for Alibaba IoT’s ecosystem, which enables users to create complete IoT nodes and gateway solutions with better time to market.

Announced last year, AliOS Things is a light-weight embedded operating system for IoT, developed by Alibaba. The company recently announced the release of AliOS Things v1.2, which includes a sensor-based component called uData. The ST sensors that have passed the AliOS validation have been integrated in uData. The two companies are cooperating on the development of IoT systems that aim to improve end-user experiences.

The LSM6DSL (shown) is a system-in-package featuring a 3D digital accelerometer and a 3D digital gyroscope that operates at 0.65 mA in high-performance mode and enables always-on low-power features for an optimal motion experience for the consumer. High robustness to mechanical shock makes the LSM6DSL well suited for the creation and manufacturing of reliable products. The LSM6DSL supports main OS requirements, offering real, virtual and batch sensors with 4 KB for dynamic data batching.

STMicroelectronics has announced the validation of its LSM6DSL 6-axis inertial sensor and LPS22HB pressure sensor for Alibaba IoT’s ecosystem, which enables users to create complete IoT (Internet-of-Things) nodes and gateway solutions with better time to market.

The LPS22HB is an ultra-compact piezoresistive absolute pressure sensor that functions as a digital output barometer. Dust-free and water-resistant by design, the sensor enables high accuracy and low-power operation. It is available in full-mold package with silicon cap and six 20µm holes guaranteeing sensor moisture resistance, relative accuracy of pressure measurement 0.1 mbar, and very low power consumption (12 µA in low-noise mode).

STMicroelectronics |

Q&A: Networking Expert Dru Lavigne

Dru Lavigne wasn’t always interested in networking applications. I recently interviewed her about how she discovered UNIX and launched her career as an OS specialist and technical writer. She also described her “to-do” list, which includes more writing, and her hopes for the future of the BSD OS.—Nan Price, Associate Editor


Dru LavigneNAN: What is your current occupation?

DRU: I’m the lead tech writer for iXsystems, a hardware solutions provider and corporate sponsor of the FreeNAS and PC-BSD open-source projects. Since both of these projects publish a comprehensive user’s guide with each software release, most of my time is spent making sure each guide is kept up to date as changes are made to these OSes. I’m also involved in the FreeBSD Documentation Project and I am currently assisting in updating and preparing the FreeBSD Handbook for publication in a two-volume format.

NAN: What is the FreeBSD Foundation?

DRU: The FreeBSD Foundation is a 501.c3 nonprofit that provides financial support and a legal entity for the FreeBSD Project.

The FreeBSD Foundation provides grants so developers can attend conferences and developer summits, sponsors developers to work on specific software projects that would benefit the FreeBSD community, interacts with companies that use FreeBSD to determine their needs, and assists in introducing developers to the community. As a director, I assist in fundraising and advocacy, reviewing project proposals, and developing relationships.

NAN: What is BSD? What is the difference between BSD and Linux?

DRU: BSD is a UNIX-like OS that was originally developed at the University of California Berkeley in the 1970s. When the university stopped developing the OS, several open-source projects began to continue development.

Its lineage differs from Linux as Linux is derived from a different UNIX branch known as SysV. Traditionally, the most noticeable difference is that SysV systems use run levels whereas BSD systems do not. The release engineering process also differs between BSD and Linux. BSD projects release an entire OS with a set of base tools included in the OS’s userland. The entire OS has a release engineering team that is responsible for the release and a security team that is responsible for security advisories until a release reaches its end-of-life (EOL). In contrast, Linux itself is only the kernel. Each distro integrates that kernel into its installer, package management system, and userland to create a complete OS.

NAN: How long have you been using BSD? When and how did you become interested?

DRU: I started using FreeBSD in 1997. I went “cold turkey” by installing it on my only computer and learned how to do what I needed to do as I needed to do it. Once I was comfortable with FreeBSD, I ventured into learning how to use NetBSD and OpenBSD, and when PC-BSD came along, I switched to that as my main desktop system.

NAN: Describe your involvement with the BSD Certification Group.

DRU: I founded the BSD Certification Group to create a community-based and psychometrically valid certification exam for system administrators of BSD OSes. The group is composed of volunteers who have been involved in BSD for quite some time as educators, authors, and/or system administrators. We have worked hard to provide a globally affordable examination that provides real value to employers.

NAN: You’ve written several books, including BSD Hacks, The Best of FreeBSD Basics and The Definitive Guide to PC-BSD. What can readers expect to learn from the books?

DRU: How to be comfortable on a UNIX system and how to think using the logic of a UNIX system.

NAN: Do you consider your books introductory or are they written for more experienced engineers?

DRU: These books are written in the style: “Now that you have BSD, did you know that you can do these cool things?” I’m a hands-on person and I like to know what I can do and to understand what I’m seeing when something I do acts differently than I expected it to.

The great thing about UNIX is that you can learn how to do something useful now, even if you have never seen a UNIX command line before. And, even if you’ve been around forever, there is always something you haven’t come across before or a cool new way to do something that you haven’t thought of before. So, these books can appeal to both the introductory user (the main target audience) as well as the advanced user (who will still pick up a trick or two before passing the book along to an introductory user).

NAN: Are you currently working on or planning any books or projects?

DRU: I do have a to-do list, book-wise. It’s interesting that I currently write the equivalent of three 300ish page books per year, but these are available for free online at  and

In addition, my current big project is the two-volume set for the FreeBSD Handbook, which will be a good 900 pages when it is complete. Once that project is finished, next in line is modernizing The Best of FreeBSD Basics for FreeBSD 10.x. Then, I’d like to write a second BSD Hacks-type book.

NAN: What do you consider to be the “next big thing” in the industry?

DRU: Since my expertise is in BSD, I’ll frame my answer from that perspective.
The first is creating usable frameworks for securing/sandboxing existing non-secure applications. FreeBSD is leading the development and research in this area in its Capsicum framework (see the article “Capsicum: Practical Capabilities for UNIX” on the University of Cambridge website).

The second is modern file systems that aren’t limited by the hardware restrictions that were around when most file systems were created. Examples include the OpenZFS storage platform and DragonFly BSD’s HAMMER file system.

NAN: Give us some background information. Where are you located? Where and what did you study?

DRU: I’m a recent transplant to Northwest Arkansas, having lived in Canada for many years. I went back to school in my early 30s to get a technical diploma in Networking and Telecommunications. I also earned the following certifications: MCSE, CNE, CCNA, CCSA, Security+, and probably others, which I have since forgotten.

NAN: How did you become interested in OSes and IT?

DRU: I was working in a dead-end position for a municipal department (low pay, very low glass ceiling) and wanted to expand my horizons. Many of our clients were being referred to a technical college for a networking program at a time when networking was a “hot” topic.

I had no idea what networking was, but figured it couldn’t be any worse than what I was doing, so I negotiated half days with my employer so I could attend classes. I quickly found that the course interested me and I seemed to be good at it.

Toward the end of the program, when I was researching employment opportunities, I noticed that the interesting and well-paying positions wanted UNIX experience. Having no idea what that was, and having no money as a poor student, I did an Internet search for “free UNIX.” The first hit was I went to the website and my gut told me “this is it.” The rest, as they say, is history.