To select an instance type, you must know their key features. Knowing the features of each instance type allows you to select a variety of instance types to manage scenarios such as insufficient resources, instance families phase-out, and use of preemptible instances and take full advantages of the elasticity and flexibility of ECS.

This topic describes how to select enterprise-level instance families, not entry-level instance families (also called shared instance families). For more information about the selection of entry-level instance families, see Overview.

Learn instance families

When you create an ECS instance, you want to select the most cost-effective and stable instance type to suit your performance, prices, and workload needs. Alibaba Cloud provides multiple instance families that are suited for different business scenarios and have different specifications of vCPUs, memory, network performance, and throughput. An instance family is further divided into multiple instance types. Instance families names are in the ecs. <Instance family> format, and instance type names are in the ecs. <Instance family>.<nx>large format.
  • ecs: product code of ECS.
  • <Instance family>: consists of lowercase letters and digits.
    • Lowercase letters are abbreviations of the performance field of the instance family. Some of the abbreviations are described as follows:
      • c: compute optimized (computational)
      • g: general purpose (general)
      • r: memory optimized (RAM)
      • ne: enhanced network performance (network enhanced)
    • Digits identify the release times of instance types within the same instance type families. A larger digit represents a newer generation of the instance family and is cheaper, more effective, and higher performing.
  • <nx>large: The number of vCPUs of the instance type. The larger the digit n is, the more vCPUs the instance type has.
For example, ecs.g6.2xlarge represents an instance type that belongs to the general purpose instance family g6 and has eight vCPUs. g6 is a newer-generation instance family than g5, g4, and sn2ne.

Select instance types

You can use one of the following methods to view the details about instance families and types to select an appropriate instance type:
  • Instance families: See this topic to learn details of instance families without logging on to an Alibaba Cloud account.
  • DescribeInstanceTypes: Call this API operation to obtain the latest performance specifications of instance types. You must log on to an Alibaba Cloud account to perform this operation. 
    aliyun ecs DescribeInstanceTypes --InstanceTypeFamily ecs.g6
  • Pricing: View the pricing information, latest special offers, and the estimated cost on the pricing page.
  • Custom Launch page in ECS console: In the Instance Types configuration option at the Basic Configuration page, you can learn more about buying guides.

Select instance families based on scenarios

The following figure shows some ECS instance families and the business scenarios for which they are suited.Compute optimized instance families

Select instance families based on applications

If you are using software or applications similar to those in the following figure, select corresponding instance families from the right side of the picture.Select instances based on workloads

Recommended instance families for scenarios such as common scenarios, game servers, and live streaming

These scenarios are CPU-compute intensive. We recommend that you select an instance type with a relatively balanced CPU-to-memory ratio, typically 1:2. Use an ultra disk as the system disk, and standard SSDs or enhanced SSDs (ESSDs) as data disks. For scenarios that require higher network performance such as on-screen video comments, you can select an instance type with higher specifications to improve the packet forwarding rate.
Scenario category Specific scenario Recommended instance family Performance requirement CPU-to-memory ratio
Common scenarios Balanced performance applications and backend applications g series, such as g6 Medium clock speed, compute-intensive 1:4
Applications with high packet forwarding rates g series, such as g6 High packet forwarding rates, compute-intensive 1:4
High-performance computation c series, such as c6 High clock speed, compute-intensive 1:2
Game applications High-performance client games c series, such as c6 High clock speed 1:2
Mobile or web games g series, such as g6 Medium clock speed 1:4
Live streaming applications Video forwarding g series, such as g6 Medium clock speed, compute-intensive 1:4
On-screen video comments g series, such as g6 High packet forwarding rates, compute-intensive 1:4

Recommended instance families for big data scenarios such as Hadoop, Spark, and Kafka

These scenarios have different performance requirements for each node. Therefore, you must balance the performance of each node, including computing performance, storage throughput, and network performance.
  • Management nodes: Select an instance family in the same manner as you would in common scenarios. For more information, see the Recommended instance families for scenarios such as common scenarios, game servers, and live streaming section.
  • Compute nodes: Select an instance family in the same manner as you would in common scenarios. For more information, see the Recommended instance families for scenarios such as common scenarios, game servers, and live streaming section. You must select instance types based on the cluster size. For example, you can select ecs.g6.4xlarge for clusters with less than 100 nodes and ecs.g6.8xlarge for clusters with more than 100 nodes.
    Note Compute nodes can be billed on preemptible instances to improve cost-effectiveness. For more information, see Overview.
  • Data nodes: require high storage throughput, high network throughput, and balanced CPU-to-memory ratio. We recommend that you use the d series big data instance family. For example, you can select ecs.d1ne.6xlarge for MapReduce or Hive and ecs.d1ne.8xlarge for Spark or MLlib.
Select instance families from big data instance families

Recommended instance families for scenarios such as databases, caches, and search scenarios

These scenarios typically require that the CPU-to-memory ratio is greater than 1:4. Some software in these scenarios is sensitive to latency and storage read/write I/O capabilities. We recommend that you select instance families whose unit price of memory is more reasonable.
Scenario category Specific scenario Recommended instance family CPU-to-memory ratio Data disk
Relational databases High performance and dependent on high availability in the application layer i series 1:4 Local SSDs, ultra disks, and standard SSDs
Small and medium databases g series, or other instance families with CPU-to-memory ratio of 1:4 1:4 Ultra disks and standard SSDs
High-performance databases r series 1:8 Ultra disks and standard SSDs
Distributed caches Medium memory usage scenarios g series, or other instance families with CPU-to-memory ratio of 1:4 1:4 Ultra disks and standard SSDs
High memory usage scenarios r series 1:8 Ultra disks and standard SSDs
NoSQL databases High performance and high availability in the application layer i series 1:4 Local SSDs, ultra disks, and standard SSDs
Small and medium databases g series, or other instance families with CPU-to-memory ratio of 1:4 1:4 Ultra disks and standard SSDs
High-performance databases r series 1:8 Ultra disks and standard SSDs
Elasticsearch Small clusters that rely on disks to ensure high data availability g series, or other instance families with CPU-to-memory ratio of 1:4 1:4 Ultra disks and standard SSDs
Large clusters and dependent on high availability in the application layer i series 1:4 Local SSDs, ultra disks, and standard SSDs
Take databases as an example. Traditionally, business systems directly connect to OLTP databases and data redundancy is mainly achieved by RAID. However, if you use ECS provided by Alibaba Cloud, you can flexibly deploy both lightly and heavily loaded databases.
  • Lightly loaded databases: Use enterprise-level instance families with disks, which are more cost-effective.
  • Heavily load databases: require high storage IOPS and low read/write latency. We recommend that you use i series instance families with local SSDs. The local SSDs are high I/O NVMe SSDs that can meet the requirements of large heavily-loaded databases.
Select instance families based on databases

Recommended instance families for scenarios such as deep learning and image processing

In these scenarios, applications require high-performance GPU accelerators. The following section describes recommendations for GPU-to-CPU ratio:

  • Deep learning training: The recommended GPU-to-CPU ratio is 1:8 to 1:12.
  • General-purpose deep learning: The recommended GPU-to-CPU ratio is 1:4 to 1:48.
  • Image recognition and reasoning: The recommended GPU-to-CPU ratio is 1:4 to 1:12.
  • Speech recognition and synthesis reasoning: The recommended GPU-to-CPU ratio is 1:16 to 1:48.

Check and adjust your select

After you select an instance type and start using the instance, we recommend that you check whether the instance type is suitable based on the performance monitoring information.

If you select the ecs.g6.xlarge instance type and find the CPU utilization is low, we recommend that you log on to the instance to check whether the memory usage is high. If the memory usage is high, you can change to another instance family with a more suitable CPU-to-memory ratio. For more information, see:

References

For more information about scenarios, see Selection of enterprise-level instance types.