Open source Euler openEuler 23.03 innovative version was officially released, the version code totalled 730 million lines, compared to openEuler 22.09, 55 million lines of new code.
The new code mainly focuses on technical innovations and basic package upgrades such as service grid data surface Kmesh acceleration, open Gemini timing database, iSulad support for mirror RO data management directory splitting, etc. Among them, 30,000 lines of original kernel code were added, mainly focusing on new features such as scheduling system and memory management system optimization of Linux 6.1 kernel.
openEuler 23.03 adopts Linux Kernel 6.1 kernel, which is an advanced technology exploration for future openEuler long life cycle version with 6.x kernel, and also facilitates developers to carry out hardware adaptation, basic technology innovation and upper layer application innovation. OpenEuler 23.03 currently supports mainstream chip architectures and hardware such as X86, Arm and Raspberry Pi. architecture and hardware.
The openEuler kernel integrates many new features of Linux Kernel 6.1, including: the scheduling subsystem supports CFS burstable bandwidth controller, improved SCHED_IDLE scheduling policy, optimized NUMA load balancing, etc. to improve system latency and batch performance; improved memory management subsystem, introducing new algorithms for identifying The memory management subsystem is improved and a new algorithm is introduced to identify hot and cold pages in NUMA nodes to optimize the distribution of hot and cold pages in the system, enhance access efficiency, and improve system performance; the XFS log synchronization mechanism, the lock-free lookup mechanism of buffer cache, and the asynchronous buffer writing mechanism are optimized to significantly improve the performance and scalability of XFS; in terms of kernel ease of use, BPF programs are compiled once and run everywhere, solving the problem of difficult porting of BPF programs; the PSI The new kernel adds support and optimization for per-cgroup PSI and IRQ/SoftIRQ PSI, which helps business to make more accurate evaluation of system resource usage and improve and optimize resource scheduling strategies.
The GCC plug-in module supports LTO complex optimization plug-ins and adds 14 new types of GIMPLE support, increasing GIMPLE coverage from 26% to 80%, significantly extending the scope of GCC plug-in support, helping developers develop efficient optimization plug-ins for different hardware platforms, and giving full play to openEuler’s diverse computing power.
secGear confidential computing framework supports Kunpeng TEE advanced interface, which supports confidential computing to build data end-to-end protection capability by shielding the underlying remote proof and key negotiation capabilities and realizes the full lifecycle security of user keys in Gaussian fully cryptographic database scenarios.
NFS client multi-path feature supports a single NFS mount point using multiple IPs for access and supports IO transmission in multiple links by proposing multiple links between client and server under a single mount point, thus improving the performance of a single mount point, while periodically detecting link status information and supporting fast switchover of link-failed IO.
openEuler is a full-scene open-source operating system for digital infrastructure, supporting four scenarios: server, cloud computing, edge, and embedded, with a continuous commitment to collaborative scene innovation.
In the embedded scenario, developers from the Institute of Software of the Chinese Academy of Sciences and Huawei introduced Mate-ROS in openEuler 23.03, with complete support for ROS2 core package and related dependencies. OpenEuler now has a minimal robot system collection and can support navigation applications in SLAM scenarios; developers from Guangdong-Hong Kong-Macao Bay Area (Guangdong) National Innovation Center have realized openEuler+UniProton @Hi3093, openEuler+RT-Thread@RK3568j hybrid key deployment; developers from Chengdu Elite introduced JailHouse for openEuler Embedded to realize embedded virtualization function, and the hybrid key deployment supports peripheral partition management;
In the cloud-native scenario, the new Kmesh, a high-performance service grid data surface built on the programmable kernel, improves inter-service communication performance by 5 times and reduces instant messaging service latency by 1 time. iSulad supports mirror RO data management directory splitting, which facilitates users’ flexible management of container and mirror data and provides technical possibilities for mirror data sub-deployment sharing.
New support for openGemini temporal database, through MPP massively parallel processing hierarchical architecture, and support for horizontal scaling, in order to cope with the huge amount of temporal data to the database brings huge challenges of performance and storage costs, the future will be deeply combined with openEuler to create an integrated platform for data storage and analysis for IoT & O&M monitoring.
In the distributed collaborative scenario, the distributed soft bus adds new transmission file and stream interfaces and introduces nstack and fillp protocols to significantly improve data transmission efficiency in some typical scenarios such as unstable networks.
The openEuler 23.03 version has 908 developers and more than 800 vendors participating in the contribution.
