Welcome to session one, auto-scaling introduction. Let's start by looking at the problem that auto-scaling is trying to alleviate. The problem is, how can we scale services automatically to match demand? That's the core critical issue that auto-scaling tries to address. And we need to worry about this because there are potentially multiple traffic requests whose loads are changing day by day or even hour by hour. Servicing applications and scenarios such as video streaming, online shopping, gaming, or even ordering food online. And these are just a few examples.

Sometimes the change in load is unpredictable and sometimes we can predict it in advance. In the case of gaming, for example, most gaming takes place in the evening. In any of these cases, we have a simple problem that we need to solve, and that is we need to scale our system to meet the demand regardless of whether we can plan for it in advance or not. What we want to do is adjust our resources to match the demand curve for the number of requests that are coming into our web application or web service. We don't want to have to do this manually, we would like to do it automatically if possible. And the job of auto-scaling is to help us achieve that automation process.

So what auto-scaling does on Alibaba cloud's platform is to automatically create and release ECS instances based on rules that you as the user specify. Further, it can configure server load balancer and relational database service white lists, without any manual intervention from you. So servers are added, they automatically gain permission to access RDS instance databases, and automatically get registered with your server load balancer.

Similarly, when the instances are removed by auto-scaling, they're removed from the RDS white list and deregistered from the server load balancer. So it fully automates the process of scale out and scale in.

So what is scale out? Auto-scaling maintains a group or pool of ECS instances and scaling out means adding additional instances to that group. Scale in is just the opposite, it removes instances from the pool. Now, why would you want to do that? Well, of course, your peak demand and your average demand probably aren't the same when your system is not under heavy load, you might want to cut costs by removing unnecessary ECS instances. You can reduce the number of resources you're currently using when your demand is low. This saves you money.

Another area where auto-scaling service has value is in health checking and elastic self-healing. The auto-scaling cluster or group can monitor the ECS instances that are part of the group. And if one of them is detected to be unhealthy, meaning it fails a health check, then it can be removed from the pool and replaced by a new healthy ECS instance, which allows your service to self-heal.

So let's recap. Auto-scaling is a management service that allows users to automatically adjust elastic compute resources, according to business needs and policies. There are multiple modes. In fact, we just discussed the three modes that auto-scaling supports. There's elastic scale out in which we add additional computing resources to the pool during peak loads. There's elastic scale in where we release ECS instances as the load reduces. And then there's elastic self-healing where it detects and replaces unhealthy ECS instances.

That concludes this session. I look forward to talking to you in the next session, session two, auto-scaling core concepts.