Key Features

• iSula lightweight container solution, unified IoT, and edge and cloud computing container solutions

  • Shortens a trace chain by three levels, and the memory usage of hundreds of containers is significantly lower than that of the Docker engine.
  • Supports standard open-source container runtime interface (CRI) and open container initiative (OCI) and flexibly interconnects with multiple OCI runtimes such as runC and Kata.
  • Secure container: combines the virtualization technology and container technology to ensure better isolation of secure containers.
  • System container: supports local file system startup to implement quick deployment, and supports systemd deployment to improve user namespace isolation.

• Kunpeng acceleration engine (KAE), supporting encryption and decryption acceleration

  • Digest algorithm SM3, which supports asynchronous models.
  • Symmetric encryption algorithm SM4, which supports asynchronous models and CTR, XTS, and CBC modes.
  • Symmetric encryption algorithm AES, which supports asynchronous models and ECB, CTR, XTS, and CBC modes.
  • Asymmetric algorithm RSA, which supports asynchronous models and key sizes 1024, 2048, 3072, and 4096.
  • Key negotiation algorithm DH, which supports asynchronous models and key sizes 768, 1024, 1536, 2048, 3072, and 4096.

• A-Tune intelligent system performance optimization engine, inferring service features and configuring optimal system parameters to ensure optimal service running

• Enhancing the performance of glibc, zlib, and gzip and fully using the NEON instruction set of AArch64 to improve the basic library performance

• Kernel feature enhancement

  • Supports ARM64 kernel hot patches.
  • Numa Aware Qspinlock: reduces cache/bus conflicts across NUMA nodes.
  • Optimizes the IOVA page table lookup and release algorithms to improve the performance of the IOMMU subsystem.
  • Optimizes the implementation of CRC32 and checksum based on ARM64 instructions and pipeline features, greatly improving data verification performance.
  • Supports ARM v8.4 Memory System Resource Partitioning and Monitoring (MPAM).

• virtualization feature enhancement

  • Interruption virtualization optimization: The process for an IRQfd to inject an interrupt is optimized, greatly improving the performance of high-performance passthrough devices (sush as NICs and SSDs).
  • Memory virtualization optimization: The Kunpeng hardware feature is used to improve the memory loading speed during VM startup.
  • Storage virtualization optimization: NUMA affinity self-binding is optimized for the iSCSI module kworker to improve the I/O performance of IP SAN disks.