--- title: "Create an Auto Scaling Group" description: "Create a TIR Auto Scaling Group from a snapshot of one of your VMs, set its size, plan, disk, scaling policy and access, and watch it start." content_type: "how-to" product: "tir-auto-scaling" platform: "TIR" url: "/tir/tir-auto-scaling/how-to/create-an-auto-scaling-group/" topic: ["create"] --- # Create an Auto Scaling Group Create an Auto Scaling Group to run a set of identical VMs that all start from one snapshot of a VM. The group grows and shrinks between **Minimum VMs** and **Maximum VMs**. Auto Scaling Groups are available in Delhi and Chennai. ## Before you begin - Permission to create VMs in the project. See [TIR Access Control](/tir/tir-iam/). - A VM to copy, or a snapshot of one. The VM must be **Running** or **Stopped**, use a CPU plan, and have a plan that supports snapshots. See [Create a VM](/tir/tir-vm/how-to/create-a-vm/) and [Create a VM snapshot](/tir/tir-snapshots/how-to/create-a-vm-snapshot/). - At least one SSH key. See [Add an SSH key](/tir/ssh-keys/how-to/add-an-ssh-key/). - A security group that opens port 22. See [Create a security group](/tir/tir-security-group/how-to/create-a-security-group/). Every VM in the group is a copy of the source snapshot, with the disk contents, installed software and files that the source VM had when the snapshot was taken. The group keeps the VM as it was at that moment, so set the VM up the way you want every group member to be before you start. ## Choose the source In the side menu, go to **Compute** and select **Auto Scaling Groups**, then select **Create Auto Scaling Group**. Choose what every VM in the group is a copy of: one of your VMs, or a snapshot of one. To find either by name, use **Search by name**. - **Take a snapshot of a VM.** On **Your VMs**, pick a CPU VM that is **Running** or **Stopped** and whose plan supports snapshots. Groups run CPU-only VMs, so a GPU VM cannot be a source. - **Use a snapshot that already exists.** On **Snapshots**, pick a **Ready** snapshot of a whole CPU VM. You can also start from the snapshot itself: in the snapshots list, the snapshot's menu has **Create Auto Scaling Group**. ### Wait for the snapshot of a VM Picking a VM takes a snapshot of it first, which usually takes a minute or two. The VM keeps running, or stays stopped. Each time you pick a VM, TIR takes a new snapshot, and each one stays under **Snapshots** afterwards. When the snapshot is **Ready**, the create form opens on it by itself. You can leave the page while you wait, because the snapshot carries on without you. If it takes more than 15 minutes, you can follow it under **Snapshots** or keep waiting. If it fails, TIR gives the reason, and you can try again or choose another source. ## Fill in the form The form needs a **Ready** snapshot of a whole CPU VM that still exists. To pick another source, select **Change source**; the values you have typed stay. 1. **Auto Scaling Group Name.** TIR suggests `scale-group-` followed by the date and time. Use 1 to 128 letters, digits, underscores or hyphens. The name must be different from every other group in the project. 2. **Select Computing Resource.** Pick a plan family, such as C3, and then a **Configuration** for its vCPU, memory and disk. Groups run on CPU plans billed by the hour. Every VM in the group uses this plan, and the plan is fixed once you create the group. 3. **Storage.** Set **Disk per VM** in steps of 10 GB. The smallest size is 30 GB, or the source disk size when that is larger. The first 30 GB of every VM is free. Each VM gets its own disk, made as a copy of the snapshot, and the disk size is fixed once you create the group. 4. **Scaling.** Set **Minimum VMs** and **Maximum VMs**, then choose the **Policy**. **Usage-based**, the default, follows real-time usage. **Scheduled** changes capacity only at the times you set, regardless of usage. The group starts with the number of VMs in **Minimum VMs**, which starts at 2 and can go from 1 up to one less than **Maximum VMs**. **Maximum VMs** starts at 5 and can be at most 60. For the usage target, see [Scale on usage](/tir/tir-auto-scaling/how-to/scale-on-usage/). For times of day, see [Scale on a schedule](/tir/tir-auto-scaling/how-to/scale-on-a-schedule/). 5. **Network & Security.** Choose at least one SSH key, and a **Security Group** that opens port 22. Every VM in the group gets the keys, the security group, and its own public IP address. See [Manage SSH keys and security groups](/tir/tir-auto-scaling/how-to/manage-ssh-keys-and-security-groups/). 6. **Advanced Settings.** For a **Usage-based** group, **Scaling behavior** sets how long usage has to stay above or below the target before the group changes size. See [Scale on usage](/tir/tir-auto-scaling/how-to/scale-on-usage/). **Health & recovery** sets how the group decides that a VM is unhealthy and replaces it. See [Set up health checks](/tir/tir-auto-scaling/how-to/set-up-health-checks/). A new VM starts from the source snapshot, so keep your data outside the group. Before you send the form, check the **Estimated cost**: an hourly range from the minimum to the maximum number of VMs, with the cost of one VM split into compute and disk. For prices, see [E2E Networks pricing](https://www.e2enetworks.com/pricing). Then select **Create Auto Scaling Group**. If a field needs fixing, its message says what to change. ## Watch the group start The group's details page opens, and the group is **Creating** while TIR checks the configuration, reserves compute capacity, applies the network and security settings, creates the group, turns on auto scaling for a **Usage-based** group, and launches the first VMs. A step can take longer than usual while cloud capacity settles over a few minutes; the launch is still going. The group is **Running** when all its first VMs are ready. ## When the group shows Failed TIR checks credit and capacity once the launch starts, after you have sent the form. If the launch fails, the group stays in the list as **Failed**, and its page gives the reason and a hint: - The region is short of capacity, with "Inventory temporarily unavailable. Please try again after some time." Retry later, or start with fewer VMs. - Low credit on a prepaid account: add credit, or start with fewer VMs. - A configuration that was rejected: contact support with the group name. - Any other reason: retry. A failed launch removes the VMs it had created. To start again from the same settings, see [Retry a failed launch](/tir/tir-auto-scaling/how-to/retry-a-failed-launch/). To remove the group, see [Delete an Auto Scaling Group](/tir/tir-auto-scaling/how-to/delete-an-auto-scaling-group/). To put the group behind a load balancer, see [Connect a load balancer](/tir/tir-auto-scaling/how-to/connect-a-load-balancer/). To see how the group behaves once it runs, see [How auto scaling works](/tir/tir-auto-scaling/concepts/how-auto-scaling-works/).