Discovering OpenAtom openEuler

Versatile Linux with Eyes on the Enterprise

The versatile Linux system known as OpenAtom openEuler is equally at home in the server room, cloud, or edge. With a huge developer community and strong corporate support, openEuler is ready to step into the foreground as a leading enterprise Linux.

The name openEuler is coming up more frequently in Linux circles. This enterprise-ready Linux is a major presence in Asia, and the project has recently been connecting with more users in Europe and the Americas.

The openEuler project began as Euler-OS, an in-house system maintained by Chinese mega-vendor Huawei. The Huawei developers needed a system that would serve a number of different roles, from server room to cloud to edge to embedded systems, and they wanted something that would adapt easily to a new generation of alternative chip architectures. Eventually, to facilitate collaboration with industry partners, Huawei contributed the code to the community as an independent project.

The openEuler project continues with the goal of providing a system that adapts easily to different architectures and use cases. The aim is to offer a system that allows the developer to “develop once and adapt to many scenarios.”

The open source project officially launched in 2019, and since 2021 openEuler has been part of the Open-Atom Foundation (see the box entitled “OpenAtom”). If you’re wondering about the name, the distro is named for the mathematician and scientist Leonhard Euler, who gave his name to the mathematical constant e (base of the natural logarithm) and was known for contributing to several other fields of math and science. The rigor and diversity of Euler’s work is an inspiration for the openEuler team to stay flexible and keep a wide-angle view of the roles their distro can play in the ever-evolving IT space.

OpenAtom

The OpenAtom Foundation is a non-profit organization focused on developing and promoting the global open source community. The foundation sees its mission as an effort to enhance industry-wide cooperation and empower open source across a broad range of sectors. Flagship projects include openEuler and the IoT-focused OpenHarmony, but OpenAtom also supports several other important projects, including the openKylin desktop OS, the OpenTenBase database, and the OpenLoong humanoid robotics initiative. The organization provides an array of services for supported projects, including open governance, communication, outreach, legal assistance, funding, and technical support.

Learn more

Introducing openEuler

The openEuler project has a very large base of contributors, users, and sponsors. The project’s close collaboration with a wide range of vendors underscores the emphasis on versatility and portability. The openEuler community includes hardware companies, chip manufactures, OS vendors, service providers, ISVs, and research institutions. Some companies and organizations maintain their own Linux distros based on openEuler, which means that openEuler serves as a kind of meta-distro (similar to Debian) for many in the global Linux space (see Table 1).

The openEuler project sees itself as a full-featured alternative to Red Hat, Suse, Ubuntu, and other enterprise distros. That might sound surprising to those in the West who are less familiar with the project and its history, but a large global development community and close ties with the hardware industry put open-Euler in a unique position for addressing many of the use-case challenges surrounding industrial Linux.

According to the openEuler project, more than 2,000 companies and institutions have joined the community, and more than 20,000 contributors have participated. Contributors include major international companies such as Intel, ARM, and AMD, as well as other organizations, such as the cloud-native Nest-OS project and the Institute for Software Chinese Academy of Sciences (ISCAS). Huawei, an openEuler strategic partner, has ranked as a top contributor in several recent Linux kernel releases (Figure 1). The breadth and depth of the openEuler project is quite impressive. According to Xiong Wei, Chairperson of the openEuler committee, more than 550 new projects have been incubated within the openEuler community over the past six years. The developers believe their investment in AI, edge, embedded, cloud, and conventional enterprise settings puts them in a strong position to adapt to future digital and intelligent infrastructure (Figure 2). According to project leaders, openEuler’s open collaboration, and its broad base of contributors, are its greatest strength.

Methods

openEuler puts out a new release every six months, and every two years they release a new Long-Term Support (LTS) edition. The LTS editions receive two years of comprehensive maintenance and support, and extended support is available for an additional two years. The emphasis of the LTS versions is on stability. The regular releases (dubbed “innovation releases”) offer a chance for developers and users to try out the latest technologies.

The openEuler download page offers an indication of the project’s versatility and scope (Figure 3). Choose an architecture (x86, aarch64, ARM32, LoongArch64, RISC-V) and choose a usage scenario (Server, Edge, Cloud, Embedded, or DevStation). Your selection will lead to other choices for hardware, image size, and other factors.

openEuler doesn’t just accommodate this broad range of use cases – the community is actively engaged in developing and integrating tools for the server room, cloud, edge, and embedded space. The following sections highlight some of the advanced technologies built into openEuler.

AI

With its deep connections to the hardware industry and broad support within the developer community, openEuler is well positioned to serve as a platform for AI modeling and development. The openEuler project has made AI a priority and offers what they call “full stack” AI support.

openEuler provides ready-made container images for the pytorch, tensorflow, and MindSpore AI frameworks. Other openEuler images offer support for the CUDA and CANN development platforms. You will also find images for installing Large Language Model (LLM) applications and toolchains, including Baichuan, ChatGLM, and iFLYTEK.

openEuler’s containerized approach provides a flexible and extensible ecosystem for optimizing AI development and workloads.

openEuler also comes with the sysHAX LLM heterogeneous acceleration runtime, which enhances performance for transformer models such as deepSeek, Qwen, Baichuan, and Llama. SysHAX assigns inference tasks to CPUs to make optimum use of CPU resources in AI systems.

In addition to its ecosystem of AI frameworks and modeling tools, open-Euler also offers an intelligent Q&A platform for streamlining code generation, troubleshooting, and Q&A. openEuler Intelligence uses multi-channel Retrieval-Augmented Generation (RAG) to bring AI techniques to AI development, giving developers greater control while improving efficiency. openEuler’s AI development framework can help with:

• Data preprocessing and labeling – quickly generates Q&A data for AI model training.
• Document quality improvement – anonymizes, deduplicates, and standardizes documents for better information security and efficiency.
• Model building – fine-tunes to tailor a model for a specific task.

You can access openEuler Intelligence via the web or a shell interface.

Further down in the stack, the open-Euler environment comes with a number of advanced components for supporting the data-intensive and compute-intensive systems that run AI applications (as described in the next section).

Server

The server room setting has undergone significant changes in recent years, and openEuler comes with several technologies designed to add efficiency and improve TCO. One example is DPUDirect, a cross-host collaboration framework that allows services to migrate between hosts for better flexibility and failover in a server room environment. DPUDirect responds to the expanding use of Data Processing Units (DPUs). DPUs, which are frequently used in AI and other data-intensive operations, combine features of a CPU and network interface to offload network-intensive tasks, improving performance in settings where complex computations are shared among multiple processors. DPUDirect facilitates offloading of the container or virtualization management plane, eliminating redundancy and improving efficiency in DPU-based environments.

openEuler also supports Enhanced Network File System (eNFS), an advanced version of NFS designed for Network Attached Storage (NAS) environments, as well as HPCRunner, a tuning assistant that reduces deployment costs and improves optimization efficiency.

EulerFS is a file system designed to work efficiently with non-volatile dual in-line memory modules (NVDIMMs). Although other file systems, such as the EXT4 file system, can work with NVDIMMs, heavy metadata management overhead can reduce the overall performance. EulerFS employs pointer-based dual directory views to reduce metadata synchronization overhead.

Cloud

Cloud technology is increasingly important in the IT space. openEuler supports a number of components designed for delivering smoother and more efficient cloud operations. HybridSched is a full-stack solution for hybrid deployment of VMs, providing enhanced OpenStack cluster scheduling and kernel-mode basic resource isolation (Figure 4). Skylark, which is incorporated with HybridSched, is a Quality-of-Service (QoS) aware resource scheduler that improves utilization when high- and low-priority VMs are deployed together.

KubeOS is a lightweight operating system that supports containerization in cloud-native scenarios. You can manage your KubeOS environment through Kubernetes, including atomic upgrades and API-based operations for both containers and node OS components. open-Euler also supports Rubik, a hybrid container deployment engine that schedules and isolates resources to improve node resource utilization, ensuring prioritization for mission-critical services. NestOS is a cloud OS incubated in the openEuler community that offers secure management and integrates popular container engines such as Docker, Podman, and iSulad.

openEuler images are currently available directly within the AWS, Azure, and Alibaba cloud systems, and you can install openEuler as a community-provided image in the Google cloud.

Edge

Edge computing is an important topic for many organizations as the old paradigm of centralized processing gives way to exploding data volumes from self-driving cars, industrial sensors, IoT systems, and other smart devices. open-Euler responds to the edge challenge by integrating KubeEdge, an open source management and provisioning tool incubated by the Cloud Native Computing Foundation. openEuler Edge integrates KubeEdge into a unified management platform (Figure 5) that streamlines AI deployments, implements service discovery and traffic forwarding, and improves processing and efficiency for edge devices.

Security Tools

Security is another important consideration, whether you manage a home network or a corporate colossus with 50,000 nodes. openEuler users can access all the powerful security components that populate the enterprise Linux space, and the openEuler developers highlight a few additional components that have received special attention.

The Integrity Management Architecture (IMA) is a mandatory access control subsystem built into Linux kernel 2.6 or later. IMA provides a deep level of file integrity protection, ensuring that boot files, system files, and other critical file assets are not subject to unauthorized change. IMA digest lists create a chain of trust that ensures processes operate without compromise.

SysMaster is an ultra-light, fault-monitoring, self-recovering replacement for the conventional init process that improves reliability and service availability without the complexity and long learning curve of systemd. open-Euler also provides support for secGear, a secure application development kit, and secPaver, a tool for developing SELinux mandatory access control policies.

Desktop

When you first install openEuler, it lands you at the command prompt. The openEuler OS is used in a variety of server settings and is designed for efficiency, without assuming the need for a full desktop environment. If you are setting up openEuler on a user workstation, or if you prefer a GUI environment to the command prompt, the openEuler documentation describes the steps for installing a desktop environment. openEuler supports the Gnome, Xfce, DDE, and UKUI desktops.

Devices and Things

OpenHarmony is an OS used with embedded systems, smartphones, tablets, Internet of Things devices, and other systems. The OpenHarmony OS was also developed by Huawei and is maintained by the OpenAtom Foundation. openEuler is closely integrated with OpenHarmony (as shown in Figure 2), offering seamless compatibility with a wide range of popular devices. The openEuler community also supports a number of other tools and systems for embedded development, as well as tools for interfacing with smartphones and other smart devices. The community even maintains a version of openEuler that runs on Raspberry Pi systems.

Conclusion

openEuler might seem like the “new kid” in some Western circles, but this expansive enterprise system is supported by thousands of developers. The openEuler environment is home to a wide range of tools for AI, edge, HPC, embedded, and conventional server room operations. Recent licensing changes at Red Hat offer an opening for other enterprise systems, and openEuler is gaining more attention as a leading alternative. Expect to hear more from this versatile Linux as new technologies such as AI and edge computing enter the foreground.

This article was made possible by support from openEuler by OpenAtom Foundation through Linux New Media's Topic Subsidy Program (https://www.linuxnewmedia.com/Topic_Subsidy).