Huawei Cloud Has Released Six Computing Instances (Three Types)
Dec 15, 2017 GMT+08:00
On Dec 15, 2017, Huawei Cloud held a meeting in Beijing for internal technology sharing, which highlighted “consistency, dedication, rapidness and stability in cloud computing”. Six technical specialists from Huawei Cloud Computing and two representatives from other heavyweight partners made speeches to an audience of about a hundred assorted cloud computing professionals. During the meeting, Huawei Cloud released six instances used for the different scenarios of general, heterogeneous, and dedicated computing, satisfying more customers’ cloud service requirements.
General-Computing Instances H3, I3, and C3 Thoroughly Improve Performance
With the increasing diversification of cloud applications, more and more enterprise services are demanding better, rapider and securer cloud computing. Over ninety percent of cloud services are deployed using general-computing products, because when overall performance, prices, and more are considered they are the best options. However, such a large user base poses a huge challenge to cloud infrastructure providers and their products.
In response, Huawei Cloud has released the three types of general-computing instances in an attempt to increase computing capability in cloud application scenarios, bringing about better, rapider and securer cloud computing to customers.
H3 is a third-generation Huawei Cloud high-performance computing (HPC) instance used for HPC scenarios. Configured with an optimized Elastic vSwitch, it provides a high network forwarding performance of 5 million PPS and supports AVX 512. Its performance in HPC scenarios exhibits a 20% increase over the previous-generation product. I3 is a third-generation Huawei Cloud ultra-high-I/O computing instance. It provides up to 3.5 million input/output operations per second (IOPS), uses the Huawei-developed intelligent network interface card (iNIC), and boasts high-performance database applications for faster data read and write operations. C3 is an enhanced third-generation Huawei Cloud general-computing instance. Configured with the new V5 CPU, it improves performance by more than 60% over the previous-generation product.
H3, I3, and C3 are all third-generation Huawei Cloud general-computing instances. They have adopted the standard CPU architecture and are configured with the custom Intel Skylake V5 CPU. Their integer computing performance and floating point computing performance are 15.95% and 15.01% greater than similar products in the industry, respectively. These third-generation computing instances are configured with the Huawei-developed Hisilicon-chip iNIC, providing customers with a super-high bandwidth of up to 50 Gbit/s and super-low latency networks. Moreover, they offer the following new HPC experiences:
l 0.8 million IOPS on a single disk
l 88μs read response
l Fast and secure erasing at 1TB/min
Heterogeneous Computing Instances PI1 and P2 Assist Huawei Cloud Computing Achieve an Amazing AI Show
The incessant evolution of AI keeps pushing application demands higher and higher for around-the-clock computing and peak data management, while at the same time demanding an increase in computing capability for cloud services.
In response, Huawei Cloud has released PI1 and P2, two heterogeneous-computing instances based on the Huawei-developed intelligent computing platform Atlas. They are designed for deep learning (DL) training and reasoning. PI1 is a dedicated instance with optimized AI reasoning. Configured with the P4 GPU, it can perform real-time code conversion and reasoning on 35 video streams simultaneously and reduces program latency by more than 93%, maximizing the performance of the AI reasoning service. P2 is a computing instance with a super-powerful AI training ability and is configured with the V100 GPU. Its AI training performance is three times greater than the P100 GPU. In addition, P2 is configured with the Huawei-developed NVMe SSD to achieve ultra-high IOPS storage performance, avoiding the cask effect.
Flagship-level Dedicated Computing Instance KunLun 9016 Helps Onboard Large-Scale Applications to the Cloud
In order to cater to the core demands of enterprises in SAP HANA, Oracle RAC, or HPC scenarios, Huawei Cloud has also released KunLun 9016. This flagship-level dedicated-computing BMS instance uses the Huawei-developed open architecture. It contains 16 processing units with each unit containing a 24-core CPU, resulting in 384 cores of physical CPUs and an ultra-large memory of 16 TB. It is also configured with the Huawei-developed SSD all-flash local storage, performing three times faster than the enterprise standard. In addition, itsupports the following functions:
l Active fault analysis
l Hot-swappable core components
l High-speed interconnections among 16 CPUs
l Hard zone technology
l Service provisioning within five minutes
Bao Liang, the director of Huawei Cloud Computing, stated, “Cloud computing is a systematic and complex project. All of us need to cooperate, combine software with hardware, and keep accumulating experience. Only in this manner will we be able to build a stable, reliable, high-speed, and secure cloud computing infrastructure. The rise of Huawei Cloud has resulted from Huawei’s continuous accumulation of underlying technologies over the years. We have repeatedly achieved technical breakthroughs and launched many patented products. Now, Huawei Cloud hopes wholeheartedly to embrace the cloud computing era by providing a stable, reliable, secure, and sustainable cloud computing infrastructure. We will improve whenever and wherever possible, enhancing computing capabilities, meeting customers’ increasing demands, and helping them achieve service innovations.”
Bao Liang, the director of Huawei Cloud Computing