Alibaba Cloud Releases Third Generation Shenlong Architecture: 5x Performance Improvement, Unified IaaS Platform

On September 26th, at the Yunqi Conference, Alibaba Cloud officially released the third-generation self-developed Shenlong architecture, fully supporting ECS ​​virtual machines, bare metal, cloud native containers, etc., running through the entire IaaS computing platform, and in terms of IOPS, PPS, etc. With up to 5x performance, users can get 100% of the computing power of a physical machine on the cloud.

Virtualization brings about a revolution in cloud computing. Its greatest value is to bring flexibility to computing, but at the same time it brings the burden of performance loss, but if the virtualization loss is not solved, the cloud computing industry will soon I encountered a technical ceiling.

After three stages of software virtualization, general hardware virtualization, and dedicated hardware chip virtualization, the third-generation Shenlong architecture realizes the unified architecture of bare metal servers, ECS virtual machines and other computing platforms, and users can obtain high-quality flexible resources. .

In addition, Alibaba Cloud also released a sixth-generation ECS example based on the Shenlong architecture, with a 20% increase in computing performance, a 30% reduction in memory latency, and a 70% reduction in ESSD latency. In addition, the highest price reduction for each product based on the new architecture is over 58%.

The Alibaba Cloud Dragon Architecture is highly adaptable to products such as containers under the cloud. Accelerated by I/O offload chip, efficient scheduling and automated flexible telescopic containerized products help users reduce computing costs by 50%, and performance is superior to physical machines.

Alibaba Cloud's millions of servers around the world will be fully upgraded to the third generation Shenlong architecture. Jiang Jiangwei, general manager of Alibaba Cloud Intelligent Basic Products Division, revealed that within Alibaba, the Shenlong architecture has been widely used in Taobao, Tmall, and rookie businesses to solve high-peak performance bottlenecks.