Huawei Cloud releases three types of six computing instances to build stronger cloud computing capabilities

On December 15, Huawei Cloud held its latest private technology sharing meeting in Beijing. The theme of this event was "Cloud Computing: Hard, Specialized, Fast, and Stable". Six Huawei cloud computing domain technology experts and two important partners spoke on stage, attracting more than 100 technology experts to attend. At the meeting, Huawei Cloud officially released six instances for three scenarios: general computing, heterogeneous computing, and dedicated computing, to meet the cloud application needs of more and wider customers.

General-purpose computing instances H3, I3, and C3 achieve comprehensive performance improvements

With the diversification of cloud applications, more and more enterprise businesses have put forward higher, faster and more stable requirements for cloud computing capabilities. Based on comprehensive performance and price factors, more than 90% of cloud businesses will choose general computing products, which also poses a great challenge to the general computing products of cloud infrastructure service providers.

[[213516]]

Photos from the event

The three general-purpose computing instances released by Huawei Cloud are designed to further improve computing performance in cloud application scenarios and bring customers higher, faster, and more stable general-purpose computing capabilities.

In high-performance computing scenarios, Huawei Cloud's third-generation high-performance computing instance H3 is equipped with Huawei Cloud's optimized Elastic vSwitch, with a network forwarding performance of up to 5 million PPS and support for AVX 512. The performance of HPC scenarios is more than 20% higher than the previous generation; the third-generation ultra-high IO type I3, up to 350W IOPS, uses self-developed smart network cards, and focuses on high-performance database applications, making data reading and writing faster; the third-generation general computing enhanced C3, equipped with the new V5 CPU, has a performance improvement of more than 60% over the previous generation.

Huawei Cloud's new third-generation general-purpose computing instances H3, I3, and C3 use a standard CPU architecture and are equipped with a customized Intel Skylake V5 CPU. The integer computing performance is 15.95% higher than similar products in the industry, and the floating-point computing performance is 15.01% higher. The third-generation computing instances released this time, paired with the intelligent network card using Huawei's self-developed HiSilicon chip, can provide customers with ultra-high bandwidth and ultra-low latency networks up to 50Gbps, as well as a new high-performance computing experience such as 800,000 IOPS per disk, 88μs read response, and 1TB/min fast and secure erase.

Heterogeneous computing instances PI1 and P2 help cloud computing achieve an amazing show in the field of AI!

The continuous development of artificial intelligence technology has driven the increasing demand for all-weather computing and data peak processing, while also placing higher requirements on the computing power of cloud services.

The heterogeneous computing instances PI1 and P2 released by Huawei Cloud are based on Huawei's self-developed Atlas intelligent computing platform. They are two heterogeneous computing instances dedicated to deep learning DL training and reasoning. The AI ​​reasoning optimization dedicated instance PI1, GPU card P4 can simultaneously perform real-time transcoding and reasoning of 35 high-definition video streams, and the program latency is reduced by more than 15 times, which can maximize the capabilities of AI reasoning business. The AI ​​super training computing instance P2, equipped with GPU card V100, has more than 3 times the performance of P100 in the field of AI training. At the same time, it is equipped with Huawei's self-developed NVMe SSD to achieve ultra-high IOPS storage performance, avoiding the short board effect, making the P2 instance achieve excellent performance in AI training.

KunLun 9016, a flagship instance dedicated to computing, helps large-scale applications migrate to the cloud

In addition, in response to the core needs of enterprises in scenarios such as SAP HANA, Oracle RAC or high-performance computing (HPC), Huawei Cloud also released the BMS computing instance KunLun 9016, a flagship product dedicated to the computing industry. KunLun 9016 adopts Huawei's self-developed open architecture, with 16 processor units, each processor with a 24-core CPU, a total of 384 physical CPU cores and 16T of ultra-large memory, with self-developed SSD all-flash local storage, and a performance improvement of 3 times compared to enterprise standard specifications. At the same time, it can provide powerful functions such as active fault analysis, online replacement of core components, 16-way CPU high-speed interconnection, hard partitioning technology, and 5-minute fast provisioning.

Bao Liang, Director of Huawei Cloud Computing Services, said: "Cloud computing is a systematic and complex project. Only by cooperating at all levels and combining software and hardware, and through continuous accumulation, can a stable, reliable, high-speed and secure cloud computing infrastructure be built. The rise of Huawei Cloud is based on the accumulation of Huawei's underlying technologies over the years, and it has continuously broken through technical bottlenecks and achieved a number of patented products. Today, Huawei Cloud is full of sincerity and aims to provide a stable, reliable, secure, and sustainable cloud computing infrastructure platform in the cloud computing era. It will continue to iterate itself, create stronger computing capabilities, meet more customer needs, and help customers achieve business innovation."

[[213517]]

Bao Liang, Director of Huawei Cloud Computing Services