×
Community Blog Maintaining Availability With Auto Scaling – Part 2

Maintaining Availability With Auto Scaling – Part 2

Part 2 of this 4-part series introduces all the aspects of Auto Scaling, the benefits and features, and formulates a wider implementation strategy.

By Shantanu Kaushik

In Part 1, we discussed different real-world organizations ranging from start-ups to multi-national enterprises and how they effectively and efficiently handle traffic loads using Auto Scaling. Alibaba Cloud introduced Auto Scaling to work through traffic load fluctuations and provide organizations with an efficient way to scale their compute resources. This accommodates the changes in network loads, which saves time, energy, and money.

What Is Auto Scaling?

Alibaba Cloud Auto Scaling is a management service that is suitable for any type of business or application. Auto Scaling automatically adjusts the number of elastic compute resources based on the demands of your business and the changes in network traffic load. When business loads increase, Auto Scaling automatically adds Elastic Compute Service (ECS) instances to ensure sufficient computing capabilities for a smoother and available user experience. As these loads decrease, Alibaba Cloud Auto Scaling removes these ECS instances automatically to save costs.

1

How Does It Work?

As the figure above shows, Alibaba Cloud Auto Scaling gets cloud data from the real-time network load using CloudMonitor.

Quick Tip: Alibaba Cloud CloudMonitor collects metrics related to Alibaba Cloud resources. The service can detect the availability of your service and allow you to set alarms on specific metrics. CloudMonitor lets you view and comprehend cloud resource usage, along with the status and health of your infrastructure and applications.

Auto Scaling implements a CPU usage scenario to enable scaling in or out. As a default setting, as the CPU usage load passes an 80% threshold, the Auto Scaling service triggers the scale out function to introduce more ECS instances to distribute the load. Similarly, the Auto Scaling service scales in function reduces the number of ECS instances when CPU loads fall below 30%, saving a considerable amount of money.

Auto Scaling helps you write custom policies and profiles to consider and include different metrics and compute load circumstances to engage or disengage Auto Scaling.

Scaling In

2

Alibaba Cloud Auto Scaling automatically disengages the underlying resources, such as ECS instances, to prevent resource waste and reduce costs. You can configure CloudMonitor to monitor your ECS instance usage in real-time. Alibaba Cloud Auto Scaling automatically scales in ECS resources based on the scaling rules you configure.

During scale in, Auto Scaling automatically releases ECS instances. Then, these ECS instances are removed from the backend server groups of the associated Server Load Balancer (SLB) instances and the whitelists of the associated ApsaraDB RDS instances.

Quick Tip: Server Load Balancer (SLB) is responsible for routing multiple requests and spreading them over server instances in a way that provides high-performance and enables optimized resource usage. The goal here is to ensure that all the server instances can share the load evenly without worrying about any server instance or compute instance (like ECS) taking more of the load than the others.

Scale Out

3

Alibaba Cloud Auto Scaling automatically increases underlying resources as the business loads increase above the pre-defined threshold. This enables you to maintain performance and resource availability to help users execute their application processes and ensure that resources are not overloaded.

You can configure CloudMonitor to monitor your ECS instance usage in real-time to achieve this. Depending on the collected metrics from CloudMonitor, Auto Scaling scales out ECS resources automatically based on the scaling rules you configure.

During scale out, Alibaba Cloud Auto Scaling creates ECS instances automatically. Then, these ECS instances are added to the backend server groups of the associated Server Load Balancer (SLB) instances and the whitelists of the associated ApsaraDB RDS instances.

While the scale out process is in service, the ECS instances are automatically created based on the pre-defined instance configuration information of the scaling group. The instance configuration information includes the instance type, operating system, and user data to enable quick deployment without complexities.

Features, Benefits, Common Scenarios

Alibaba Cloud Auto Scaling provides a host of features and benefits for demanding usage scenarios. A usage overview is listed below:

With Alibaba Cloud Auto Scaling, you can:

  • Add or remove ECS instances automatically
  • Add or remove ECS instances from the backend server groups of Server Load Balancer instances automatically
  • Add or remove IP addresses of ECS instances from the whitelists of ApsaraDB for RDS instances automatically
  • Auto scale the resources on-demand to respond to traffic spikes in real-time
  • Create and release ECS instances without manual intervention automatically
  • Configure SLB instances and whitelists of ApsaraDB for RDS instances automatically
  • Schedule, customize, and fix the minimum number of instances
  • Replace unhealthy instances automatically
  • Enable API operations to allow you to monitor instances using external monitoring systems
  • Schedule cloud computing resources intelligently to respond to complex scenarios

Alibaba Cloud Auto Scaling works in three scaling modes:

  1. Fixed Mode – This mode enables you to run your services with the minimum number of ECS instances required to support regular business.
  2. Scheduled Mode – This mode allows you to configure the scheduled tasks by adding or removing ECS instances for a fixed time. This mode can also be coupled with Dynamic Scaling mode.
  3. Dynamic Mode – This mode is among the most flexible modes. As the name suggests, this mode reads the load metrics from the CloudMonitor and dynamically adds or removes the ECS instances depending on load.

We will discuss Alibaba Cloud Auto Scaling scaling modes in further detail in Part 3.

Alibaba Cloud Auto Scaling works in various high-demand usage scenarios:

  • Video Streaming or Broadcasting – This use case requires the dynamic mode of scaling in and out, based on the collected metrics from CloudMonitor. Video streaming is a bandwidth-hogging service that needs Auto Scaling depending on the number of users.
  • Gaming – Gaming traffic needs low latency solutions to provide users with a smoother experience by adding resources in real-time.
  • E-Commerce – Traffic can surge tremendously during big sales. Auto Scaling worked seamlessly during the Double 11 Global Shopping Festival, which incurs millions of orders an hour.

Wrapping Up

In Part 3 of this 4-part series, we will discuss Elastic Recovery, the Auto Scaling workflow, and scaling modes.

Upcoming Articles

  1. Maintaining Availability With Auto Scaling – Part 3
  2. Maintaining Availability With Auto Scaling – Part 4
  3. Multi-Tier Approach With VPC – Part 1
  4. Multi-Tier Approach With VPC – Part 2
  5. Multi-Tier Approach With VPC – Part 3
0 0 0
Share on

Alibaba Clouder

2,530 posts | 569 followers

You may also like

Comments

Alibaba Clouder

2,530 posts | 569 followers

Related Products