The next section we will cover is ECS Instances. An ECS instance is a virtual machine that contains basic computing components such as the CPU, memory, operating system, network bandwidth, and disks. Once created, you can customize and modify the configuration of an ECS instance. For example: Add or remove additional Cloud Disks.

ECS instances are categorized into different families, based on the business needs to which those families can be applied, and an instance family also has many instance types based on different CPU and memory configurations. An ECS instance defines two basic attributes: the CPU and memory configuration.

The instance types follow a naming convention which depicts the instance family, instance generation and instance size, for Example, ecs.g5.large. ecs is a prefix (All ECS instances have this in the name), ‘g’ denotes instance family (in this case general purpose), 5 denotes the instance generation and implies the CPU to RAM ratio, in this case, a ratio of 1 to 4 (this means that for each CPU there is 4 GB RAM), and large denotes the instance size.

ecs.g5.large is the smallest instance in the general-purpose family and this instance has 2 CPUs, so with a ratio of CPU to RAM of 1 to 4 this instance has 2 CPUs and therefore 8GB of RAM.

ecs.g5.xlarge is the next in the family tree so it has 4 CPUs and 16GB of RAM.

ecs.g5.2xlarge is the next in the family tree so it has 8 CPUs and there 32GB of RAM.

There are 3 main types of families: X86-Architecture, Heterogenous Computing, and ECS Bare Metal Instances. The first family type is X86-Architecture and, within this section, there are 7 different subtypes as follows:

Entry Level (Shared Burstable): You can accumulate CPU credits for your burstable instances, and consume those credits to increase the computing power of your workloads when required. Used for Web application servers, Lightweight applications, and development and testing environments.

General Purpose: Used for Websites, application servers, Game servers, Small and medium-sized database systems.

Memory Optimised: Used for data analysis and mining, and other memory-intensive enterprise applications.

Big Data: Used for Enterprises that need to compute, store, and analyze large volumes of data.

Local SSD: Used for Online transaction processing (OLTP) and high-performance databases.

High Clock Speed: Used for on-screen video and telecom data forwarding, High-performance scientific and engineering apps.

The second family type is Heterogenous Computing, and there are 2 main subtypes as follows:

GPU-based compute-optimized: Used for Rendering and multimedia encoding and decoding, Machine learning, high-performance computing, and high-performance databases, Other server-high end workloads that require powerful concurrent floating-point compute capabilities.

Field-Programable-Gate-Array-based compute-optimized: Used for Deep learning and reasoning, Genomics research, Financial analysis, Image transcoding, Computational workloads such as real-time video processing and security management.

The last of the 3 family types is ECS Bare Metal Instances: This is a compute service that combines the elasticity of virtual machines and the performance and features of physical machines. The virtualization used by ECS Bare Metal Instances is optimized to support common ECS instances and nested virtualization. ECS Bare Metal Instances use virtualization 2.0 to provide business applications with direct access to the processor and memory resources of the underlying servers without virtualization overheads. These are ideal for applications that need to run in a non-virtualized environment.

Lectures

Overview - ECS Concepts - ECS Images - ECS Storage - ECS Snapshots - ECS Security Groups - ECS Networking - ECS Demo