Recently, Tencent announced an open source cloud operating system TencentOS its cloud server kernel (TencentOS Server Kernel, also known as Tencent Linux, referred Tlinux).
Tencent introduced the industry compared to other versions of Linux distributions, Tencent Linux very competitive in resource scheduling flexibility, container support, system performance and security levels, especially for cloud environments.
TencentOS Kernel Things Tencent cloud operating system TencentOS tiny after, TencentOS second family of external open source projects. TencentOS Tencent cloud operating system family, the main portion Tencent internet cloud infrastructure development, covering the data center, a desktop system, the edge devices and the like things terminal scenarios, cloud platform for building, access and application capabilities.
Tencent cloud operating system team, said: "Tencent cloud server operating system kernel TencentOS Kernel open source, not only can share Tencent cloud technology and experience in the server operating systems and the global developer, but also to learn from the outstanding achievements of the global operating system in the field of servers and innovative ideas, boost overall server operating system ecological prosperity. "
Look at the specific features of the project:
Dedicated resource scheduling scheme, enhanced system flexibility scalability
Traditional Linux system, fairness is the core logic of resource scheduling algorithms. Fair scheduling system can provide the greatest degree of versatility. However, this fair will result in the effective utilization of resources elastic capacity is low, the system is greatly restricted, business can not allocate resources on demand recovery, including resource quantity and quality of resources, such as resource allocation speed, ability to seize.
Tencent Linux developed a dedicated resource scheduling algorithms, significantly increasing the resources elasticity machine. Business can specify the number of resource allocation and level requirements, system services available from different levels of the same number of resources there are obvious differences.
Meanwhile, Tencent Linux-developed elastic CPU scheduling algorithms, online business mix away from cloth at the scene of return is very significant. Without affecting the quality of the online business, enhance the overall CPU utilization up to three times lower part of the business scenarios can enhance the overall CPU utilization to 90%.
Enhanced resource isolation
Safe isolation of resources has always been the core issue of container virtualization platform, isolation characteristics provided by the kernel community is far from meeting the needs of business, some basic system status information, CPU and disk status information etc. without making isolation, lower part of the scene even direct services may become unavailable.
Tencent Linux from business needs, first of all necessary for system status, such as cpuinfo, stat, loadavg, meminfo, vmstat, diskstats such as isolation and uptime enhanced to ensure the application container can get the correct system status information.
More importantly, the system also provides other features including NVME IO isolation, completely solve the IO control group in a multi-device context using the low resource queue, does not support isolation in proportion and so on, to ensure that the IO isolation under different scenarios.
System Security and Performance Optimization
Industry kernel patching is mainly developed for the X86 architecture, such as lack of support for ARM64 architecture. Tencent Linux by implementing class FMENTRY, FTRACE with REGS function, increased ARM64 support to KPATCH and other implements in the kernel hot patching program for ARM64 architecture.
Meanwhile, Tencent Linux provides process GDB disable function to prevent cross-process acquisition memory, dynamic load libraries, business processes to ensure data security.
Performance, Tencent Linux for computing, storage and network subsystem are optimized, e.g. PAGE CACHE LIMIT function, the page cache usage limit, the available memory of the system as much as possible to meet the needs of the business; system further add multiple sysctl / proc control interface, the kernel boot parameters to optimize the user experience.