edit-icon download-icon

Compute

Last Updated: Sep 06, 2018

Alibaba Cloud for AWS Professionals

Contents

This article discusses the main differences and similarities between AWS and Alibaba Cloud compute services. It covers the following products:

Feature AWS Alibaba Cloud
Virtual Servers Elastic Compute Cloud (EC2) Elastic Compute Service (ECS)
Block Storage EBS ECS Disk
Automatic Scaling Auto Scaling Auto Scaling
Container Service EC2 Container Service (ECS) Container Service
High Performance Computing High Performance Computing (HPC) Elastic High Performance Computing (E-HPC)

1. Virtual servers

Both Amzone EC2 and Alibaba Cloud ECS provide virtual servers for cloud computing. Virtual servers, or virtual machines, provide IaaS services to users. Alibaba Cloud and AWS servers share similar terminologies and concepts, as shown in the following table:

Feature Amazon EC2 Alibaba Cloud ECS
Virtual machine Instance Instance
Images Amazon Machine Image Images
Temporary Instance Type Spot instance Spot instance(Preemptible instance)
Firewall Security Group Security Group
Automatic Instance Scaling Auto Scaling Auto Scaling
Persistent Block Storage of Instances Elastic Block Store Cloud Disk
Local Mount Disk Instance storage Local disk
Shared Block Storage N/A Shared Block Storage
Disk Volume Backup Snapshot Snapshot
VM Import RAW, OVA, VMDK, and VHD RAW, VHD. QCOW2, VMDK, VDI, VHD (vpc), VHDX, qcow1 or QED format images can be converted to VHD or RAW format images for import.
Deployment Location Zone Zone

1.1 Login instance

AWS and Alibaba Cloud allows you to connect to your virtual server through SSH protocol. Alibaba Cloud also allows you to connect directly to the server using the Management Terminal on the console.

Connecting through SSH protocol: Alibaba Cloud ECS and Amazon EC2 differ in login instance methods. Though both servers provide SSH keys for login, Alibaba Cloud allows an SSH key to be created after an instance startup is successful and a login after the instance is bound. Furthermore, Alibaba Cloud ECS provides the username + password login method for users who are not familiar with SSH keys.

Connecting by Management Terminal: Besides the method of connecting to your virtual machine by SSH client tools, Alibaba Cloud provides an easy way to allow users to connect to ECS directly by Management Terminal (also called VNC) on the console. VNC connection is a better option if you are checking the boot procedure, configuring BIOS during startup, reconfiguring the firewall, or troubleshooting when the instance malfunctions.

Alibaba Cloud ECS and Amazon EC2 employ the same method to categorize VM instances by specifications and types, but the categorization differs in terms of CPU, memory, storage performance, and network capability. Amazon EC2 categorizes instances by configuration, while Alibaba Cloud ECS categorizes instances into different families by application scenarios. Each family is composed of different instance types. Learn more about Alibaba Cloud ECS instance families at Alibaba Cloud ECS Instance Families.

Alibaba Cloud ECS provides multiple types of instance families and configurations to meet business requirements and performance requirements in different scenarios. The following table lists Amazon EC2 instance types and Alibaba Cloud ECS instance families.

TargetGroup Scenario Amazon EC2 Instance Type  Alibaba Cloud ECS Instance Family
Entry Level General Type t2,t3 t5 
EnterPrise Level General type m4, m5 g5
Computing instance c4, c5 c5
High-frequency computing instance c5 c4, cm4, ce4, hfc5
Memory instance r4 r5, re4
x1 se1
Big data instance d2 d1,d1ne
Local SSD instance i2, i3 i1, i2
Instance of high capability of packet forwarding N/A sn1ne, sn2ne, se1ne
GPU visualization computing instance g2, g3 ga1
GPU computing instance p2, p3 gn4, gn5
FPGA computing instance f1 f1,f2

1.2 Instance images and Create instance

Instance image refers to the running environment template for virtual machine instances. Amazon EC2 and Alibaba Cloud ECS use images to create instances. AWS instance images are referred to as Amazon Machine Images (AMIs), and Alibaba Cloud instance images are simply referred to as Images.

When an instance is created, Alibaba Cloud ECS provides four types of images for users to choose from: public images, cloud marketplace images, user shared images, and custom images. Amazon EC2 provides official AMI templates, custom AMIs, cloud marketplace AMIs, and community AMIs.

Public images are system images provided by Alibaba Cloud ECS for users, which are similar to the AWS official AMI templates.

Cloud marketplace images are provided by third-party ISV partners on the Alibaba Cloud Marketplace. Beside the OS, cloud marketplace images may be preinstalled with other software and services.

Like the custom AMIs of AWS, Alibaba Cloud custom images are created by users based on snapshots or the current state of an instance. Custom images can be shared to other specific Alibaba Cloud users by using the image sharing function of Alibaba Cloud ECS.

The community AMIs of AWS is available to all AWS accounts, a feature which is currently not supported on Alibaba Cloud ECS.

Alibaba Cloud ECS images and Amazon EC2 AMI are both regional resources Custom images and shared images can be used only in the same region. To use the images in a different region, you need to replicate them to that region first.

Category AWS Alibaba Cloud
Basic EBS magnetic Basic cloud disk
Intermediate General SSD (gp2) Ultra cloud disk
Advanced (I/O Optimized) PIOPS (io1) SSD cloud disk

Additionally, Alibaba Cloud also provides two types of local block storage for instances, which feature low access latency, high random IOPS, and high I/O throughput: local NVMe SSD and SATA HDD. These ECS type families with local block disk are similar to Amazon EBS of local storage.In terms of block storage performance, Amazon EBS Provisioned IOPS SSD (io1) supports up to 32,000 random IOPS read and write performance on a single disk. Alibaba Cloud’s latest ESSD cloud disk provides up to 1 million random IOPS read and write performance on a single disk.

Instance pricing model: Alibaba Cloud ECS provides pay-as-you-go and yearly/monthly subscription options. Amazon EC2 only supports 1 or 3 years of RI. The pay-as-you-go model is similar to that of Amazon EC2, which is a post-paid based payment. The yearly/monthly purchase is a payment and settlement method used in the prepaid model.

Similar to Amazon EC2 Spot Instance, Alibaba Cloud ECS currently provides billing models for spot instances. For more information on Alibaba Cloud Spot instances, see Alibaba Cloud Spot instances.

Instance configuration modification: The yearly/monthly instances of Alibaba Cloud ECS support anytime upgrade and renewal for configuration downgrading, allowing users to conveniently adjust the ECS specifications according to sever loads and business requirements.

2. Automatic scaling

Auto Scaling is a feature that automatically adjusts computing resources based on the volume of user requests. Both AWS and Alibaba Cloud support automatic scaling, and the products share the same name (Auto Scaling). Auto Scaling enables users to set automatic scaling policies according to actual business circumstances and add/release ECS instance resources to meet business requirements.

Both Alibaba Cloud Auto Scaling and AWS Auto Scaling support the following scaling modes:

  • Custom mode: Add/release compute instances, such as Amazon EC2 and Alibaba Cloud ECS, manually.
  • Scheduled mode: Users configure periodic tasks to add/release compute instances according to a schedule.
  • Dynamic mode: Auto Scaling is performed automatically by monitoring compute resources. AWS adds/releases EC2 instances based on the CloudWatch scaling policy, while Alibaba Cloud adds/releases ECS instances based on the CloudMonitor scaling policy.
Function Feature Amazon Auto Scaling Alibaba Cloud Auto Scaling
Custom mode Supported Supported
Scheduled mode Supported Supported
Dynamic mode Supported Supported

AWS Auto Scaling is enabled by Amazon CloudWatch and is available for use at no additional fees. However, the usage of the Amazon EC2 instance added by Auto Scaling, and Amazon CloudWatch service fees, still apply and are billed separately.

Similar to AWS, Alibaba Cloud Auto Scaling is offered to customers at no extra cost. You will only be charged for the usage of the ECS instances automatically created or manually added to Auto Scaling.

3. Container service

Amazon Elastic Container Service (ECS) and Alibaba Cloud Container Service are container orchestration services that simplify container management and application scaling. Both services replace the need to install, operate, and scale your container cluster infrastructure.

Alibaba Cloud Container Service enables you to efficiently run and manage Docker applications on a distributed cluster of Alibaba Cloud ECS instances. Being a fully-managed service, Container Service helps you to focus on your applications rather than managing container infrastructure.

AWS ECS and Alibaba Cloud Container Service use the same service model. With Alibaba Cloud Container Service, users can deploy, manage, and expand Docker containers with ease. Alibaba Cloud Container Service supports App lifecycle management using Docker containers, provides a variety of App publishing methods and continuous delivery capabilities, supports microservice architecture, and integrates with Server Load Balancer, Security Group, Cloud Disk, and Resource Access Management.

Like Amazon Elastic Container Registry, Alibaba Cloud Container Service provides an image warehouse (Container registry) hosted by Alibaba Cloud, allowing access to official Alibaba Cloud images and those of Docker, and enables accelerated access to official Docker images.

Amazon ECS and Alibaba Cloud Container Service differ in their pricing models. Amazon ECS provides two different pricing models: Fargate Launch Type Model and EC2 Launch Type Model.

Like the second pricing model of Amazon ECS, Alibaba Cloud Container Service is free of charge. Resources used in collaboration with Container Server (including Server Load Balancer and ECS) are charged separately. ECS instances or Server Load Balancer instances automatically created from the Container Service or manually added are billed by their respective prices.

4. High performance computing

AWS High Performance Computing (HPC) and Alibaba Cloud Elastic High Performance Computing (E-HPC) are optimized compute resources created by using parallel computing and aggregating multiple computing capabilities.

AWS and Alibaba Cloud both provide high performance computing capabilities that allow users to solve complex, compute intensive challenges in the field of science, engineering, and business.

However, Alibaba Cloud E-HPC provides an all-in-one high performance computing service which we call HPCaaS. E-HPC supports Infrastructure as a Service (IaaS) with high-performance CPU and heterogeneous computing GPU instances, Platform as a Service (PaaS) with high-performance computing software stack, and Software as a Service (SaaS) with application template customization.

There are two different ways to help you deploy and manage an HPC cluster on AWS. One is using a fully-managed service offered by AWS, such as AWS Batch, Lambda, and Step Functions, while another way is by using third-party software.

Unlike AWS HPC, Alibaba Cloud E-HPC provides a fully-managed control panel that allows user to deploy an HPC cluster, manage users, upload job data, and submit the user job.

To launch or scale up HPC clusters on AWS, users can benefit from automation using AWS Auto Scaling. Alibaba Cloud E-HPC also provides auto scaling capability to allow user to scale up/down the cluster ECS nodes automatically.

Every AWS service provides encryption and options to grant granular permissions for each user while maintaining the ability to share data across approved users.

Similar to AWS HPC, Alibaba Cloud E-HPC is protected by multi-tenant security isolation of the highest level that is provided by ECS, EGS, and VPC. Furthermore, E-HPC service also allows user to manage user permissions and passwords with the E-HPC console.

AWS users only need to pay for the services they consume, and once the resources have been stopped, there are no additional costs or termination fees.

Like AWS, E-HPC is billed for the resources that you created: ECS, E-HPC, Network Attached Storage (NAS), and Internet traffic of login nodes. E-HPC is free of charge during the test invitation phase.

These two services can be compared as follows:

Function Feature Amazon HPC Alibaba Cloud E-HPC
Cluster Deployment and Management Third-party software E-HPC Console
User Management Third-party software E-HPC Console
Auto Scale Supported Supported
Secure Supported Supported
Thank you! We've received your feedback.