choose Alibaba Cloud Enterprise Server There are many different configurations and models. We need to understand the characteristics of these models and how to make full use of the elasticity of Alibaba Cloud ECS servers in different business scenarios to select ECS servers suitable for our own business. In this article, Lao Wei mainly introduces how to select Alibaba Cloud enterprise level ECS, excluding entry-level ECS such as shared and burst performance instances. Please move on to the selection of entry-level examples Shared or Burst performance instance Page.
Alibaba Cloud Instance Type Family Introduction
Alibaba Cloud provides multiple instance type families for ECS of different specifications and configurations according to the scope of application and scenarios. One instance type family includes multiple instance types. Before selecting an ECS instance, users should make a balanced choice between price performance and stability based on the comprehensive performance, price and other parameters of the server.
Alibaba Cloud ECS instance type family name format ecs< Specification family>, and the instance type name is ecs< Specification family><nx>large.
Ecs: ECS product code abbreviation.
<Specification family>: It consists of lowercase letters and numbers. The lowercase letters are word abbreviations, representing the performance scenario of the specification family.
For example, the meaning of lowercase letters is as follows:
- c: Generally refers to the calculation type;
- g: Generally, it refers to universal type;
- r: Generally, it refers to memory type;
- Ne: generally refers to network enhanced type;
Numbers generally differ from the release time of the same type of specification family. The larger number represents a new generation of specification family, with higher cost performance, low price and good performance.
<nx>large: The larger the n, the more CPU cores.
For example, ecs.g6.2xlarge represents an instance type in the generic g6 specification family, which has 8 CPU hyper threading cores. Compared with g5, g6 is a generic instance specification family of the new generation DPCA architecture, with stronger performance.
AliCloud ECS selection
To make it easier to compare instance performance when selecting, you can learn more about instance details from the following ways.
Instance Type Family : Check the document content to understand the detailed product situation of the instance type family;
Cloud product pricing page : Quickly understand the price information of ECS servers and simply estimate the cost of going to the cloud;
ECS customized purchase page : Understand the detailed parameters of CPU, memory, cloud disk, network, operating system, region, etc. on the instance configuration page;
AliCloud Cloud Website : Both new and old users can enjoy it, and they can get limited coupons for free in the cloud discount gathering place.
Choose AliCloud ECS servers according to the usage scenarios
According to specific use scenarios, including but not limited to firewall, load balancing, NAT gateway, memory database, finance, government enterprise, enterprise security, Redis application parameter server, blockchain application, video transcoding, various distributed databases, big data computing, scientific computing, large games, rendering video coding and decoding and other different types of choices.
Extended reading: Alibaba Cloud enterprise server configuration and product selection strategy
The difference and choice between Alibaba Cloud server entry level and enterprise level
Performance gap and selection between Alibaba Cloud entry-level servers and enterprise instances
As shown in the figure below, the virtual machine can be selected for general applications and general requirements. If the calculation is relatively large, the bare metal server needs to be selected.
The following figure includes the Alibaba Cloud ECS heterogeneous computing instance type family and its corresponding business scenarios. It covers deep learning test, deep learning reasoning, video transcoding, graphic image processing, high-performance computing, scientific computing, physical isolation, picture transcoding, video transcoding, encryption computing, genomics research, financial analysis, deep learning reasoning and other scenarios.
Choose AliCloud ECS based on typical applications
The software or application can select the corresponding instance type family on the right in the following figure. It includes web server, middleware, application server, cache, database, big data, AI machine learning, etc.
Self built service selects AliCloud ECS
Self built services can select the corresponding instance type family according to the application.
General scenes, game clothes, live video scenes
This scenario mainly selects CPU computing intensive. The ratio of CPU to memory is usually 1:2, efficient cloud disk, and data disk are SSD cloud disk or ESSD cloud disk. If the business needs stronger network performance, such as video bullet screen, you can choose a higher specification instance in the same series to improve the network receiving and sending capability (PPS).
Hadoop, Spark, Kafka big data scenarios
In this type of scenario, because different nodes are involved, the performance requirements are complex. You need to balance the performance of each node, including computing, storage throughput, and network performance.
For details of management node computing node and data node, please move to Recommended general scenes, game clothes and live video scenes 。
Database, cache, search scenario recommendation
This scenario requires a large amount of memory, so the ratio of CPU processor to memory is greater than 1:4. It is sensitive to the storage I/O read/write ability and timely ductility. It is recommended to select a specification family with a high cost performance ratio of memory.
Deep learning, image processing scene recommendation
Such applications require high-performance GPU accelerators. The recommended GPU models for common scenarios are shown in the following figure:
Scenarios include: natural language understanding, voice recognition, advertising recommendation, video analysis, image search, image classification, target detection, video compression, image transcoding, etc.