Docs

Elastic Compute Service

Overview

Application scenario

Web applications

General-purpose instance types balance CPU and memory performance and are suitable for most web applications. For compute-intensive or memory-intensive workloads, you can choose Compute-optimized or Memory-optimized instances.

Big data applications

The big data-based instance type offers Elastic Compute Service (ECS) instances with large storage capacity and direct access to high-throughput local SATA HDDs. This instance type is suitable for storage nodes in HDFS applications and provides a balance among capacity, computation, and memory for big data processing.

Applications with high I/O requirements

For IOPS-sensitive applications such as Redis and MySQL clusters, the platform provides Local-SSD series. This way, ECS directly accesses the local NVMe SSD. With exclusive access to a local disk, each server provides storage performance at the million IOPS level.

High-performance computing

In compute-intensive, high-throughput scenarios like scientific computation, genetic engineering, EDA, game animation, and biopharmaceuticals, the platform offers High Frequency series. These series provide high-performance computing capability and high stability. This helps you efficiently, reliably, and rapidly run large-scale applications.

AI deep learning

Model training

GPU instances equipped with V100, A100, A30 graphics cards are suitable for AI model training. The instances offer large memory and high access speed and provide high computing power for multi-card parallelism by integrating NVLink for multi-card interconnects.

Application inference

GPU instances equipped with T4, A10 graphics cards offer acceleration with a high performance-to-energy ratio in AI inference. This type is widely used in scenarios like image recognition and language translation.

Image/video rendering

Heterogeneous GPU rendering instances provide high-performing 3D graphics virtualization for efficient real-time image rendering. This type is widely used in fields such as engineering graphics, video rendering, and animation production, significantly increasing real-time and offline rendering efficiency.

Last updated: 2026.06.18 16:37:45