An enterprise-level user can select configurations by following the process. For more information about the instance types, see Instance generations and type families.
As an enterprise-lever user of ECS, you may have specific requirements. To meet your requirements, Alibaba Cloud provides you with recommendations for instance configuration in the following scenarios:
Note:
For more scenarios, see Select configurations for enterprise-level users.
Balanced performance
A balanced CPU to memory ratio is required to meet the application resource requirements in most scenarios.
Applications with high packet forwarding rate
High packet forwarding rate is required. You can select a more reasonable computing capacity to memory resource ratio based on the specific scenarios.
High performance computing
Many computing resources are required. GPU parallel computing and a high clock speed are typical applications in this scenario.
High-performance client games
Your services require a high-frequency processor to carry more users. Therefore, a high clock speed is required in this scenario.
Mobile, Web games, and forwarding videos
Many computing resources are required for this scenario. A CPU to memory ratio of 1:2 can help achieve optimal cost performance of computing resources.
Live bullet screen
High packet forwarding rate is required for this scenario. You can select a more reasonable computing capacity to memory resource ratio based on the specific scenarios.
Relational databases
In this scenario, SSD cloud disks or higher-performance NVMe SSD local disks are required to provide higher IOPS capability and a low read latency. The CPU to memory ratio is balanced (1:4) or the memory proportion is larger (1:8).
Distributed cache
In this scenario, a balanced CPU to memory ratio (1:4) or a higher memory proportion (1:8), and stable computing performance are required.
NoSQL databases
In this scenario, SSD cloud disks or higher-performance NVMe SSD local disks are required to provide higher IOPS capacity and a low read latency. The CPU to memory ratio is balanced (1:4) or the memory proportion is larger (1:8).
Elasticsearch
In this scenario, SSD cloud disks or higher-performance NVMe SSD local disks are required to provide higher IOPS capacity and a low read latency. The CPU to memory ratio is balanced (1:4) or the memory proportion is larger (1:8).
Hadoop
Data nodes require a high disk throughput, high network throughput, and balanced CPU to memory ratio. Computing nodes focus more on the computing performance, network bandwidth, and CPU to memory ratio.
Spark
Data nodes require a high disk throughput, high network throughput, and balanced CPU to memory ratio. Computing nodes focus more on the computing performance, network bandwidth, and CPU to memory ratio.
Kafka
Data nodes require a high disk throughput, high network throughput, and balanced CPU to memory ratio. Computing nodes focus more on the computing performance, network bandwidth, and CPU to memory ratio.
Machine learning
In this scenario, a high performance Nvidia GPU computing processor is required, and the memory size must be at least twice the video memory.
Video encoding
In this scenario, a high performance GPU computing processor or a high performance CPU is required for encoding and decoding.
Rendering
In this scenario, a high performance GPU computing processor is required for rendering.