# Auto Scaling

Auto-scaling automatically adjusts the number of VM instances in response to real-time demand, ensuring availability while minimizing costs.

Before you begin, you need an existing [Load Balancer](https://docs.zcp.zsoftly.ca/public-cloud/load-balancer) with a load-balancing rule that has no instances attached. An autoscaling group attaches to that rule to distribute traffic, and user-bound VMs cannot coexist with an autoscale group on the same rule.

### Create an Auto-Scaling Group

- From the left-hand menu, click Auto-Scaling.
- Click Create New.

![Create an Auto-Scaling Group form](https://docs.zcp.zsoftly.ca/_astro/auto-scaling-create-an-auto-scaling-group.LlOZfrLk_2phVvK.webp)

### Steps

1. Project: select the project.
2. Location: select the data center.
3. Network: select or create a network.
4. Load Balancer: select an existing load balancer.
5. Rule: select a load-balancing rule with no instances attached.
6. Image: select the OS (Linux or Windows) and version.
7. Storage Type: select the storage type (for example, `b2.g1` or `b2.g2`).
8. VM Autoscale Plan: select an autoscale plan.
9. Plan: select a compute plan, or set a custom plan (CPU, memory, storage).
10. Server Settings: add an SSH key and a startup script (both optional).
11. Affinity Groups: add the instances to an affinity group (optional).
12. Capacity Planner: set Minimum Instances, Maximum Instances, Expunge VM Grace Period (seconds), and Polling Interval (seconds).
13. Scale Up Policy: add at least one. Set a Name, Duration (seconds), and Quiet Time (seconds).
14. Scale Down Policy: add at least one. Set a Name, Duration (seconds), and Quiet Time (seconds).
15. Name: enter a unique name.
16. Create: select a billing cycle and click Create.