Create autoscaling at CMP
This page is in progress. Content, create-form fields, and screenshots will be expanded as CMP Autoscale documentation is completed.
This page covers how VM Autoscaling is offered and created through CMP on CloudStack — admin prerequisites and the customer create flow at a high level.
Read CloudStack considerations and practical use cases first so the application and golden template are ready before customers enable AutoScale.
Before customers can create AutoScale
Admin — Cloud Provider and packages
- Enable VM Autoscale (and Load Balancer if LB is required) in Cloud Provider Setup Wizard Step 1
- Create an Active VM Autoscale package for the zone (engine fee optional — can be
0) - Ensure Virtual Machine, Load Balancer, and IP Address packages exist for resources the group will use
- Ensure My Template is available so customers can register a golden image — typically create from VM root volume (VM must be powered off)
CloudStack / network readiness
- Network or VPC offerings support Load Balancer (and AutoScale where your CloudStack version requires it on the offering)
- Quota allows VM Autoscale (and related VM / LB / IP limits) — see Global quotas
On VPC, do not use the Source NAT IP for load balancer rules. Customers need a separately acquired public IP. See Load Balancer — VPC Source NAT.
Recommended customer sequence
1. Build & test horizontally scalable app on multiple VMs
2. Create golden VM → register Template (My Template)
3. Create / select Load Balancer (eligible public IP)
4. Create AutoScale in CMP (profile, policies, min/max, template)
5. Scale-up / scale-down runs from CloudStack; CMP bills VMs (+ optional engine fee)
1. Prepare the application and golden template
Follow CloudStack considerations: shared DB/cache, auto-start, no VM-local-only session state.
Then create a template from the golden VM — power off the VM and use Create from VM root volume. When the template is ready, it appears under Create Instance → My Templates and can be selected for the AutoScale VM profile.
2. Create a load balancer
Customers create a load balancer from the CMP Load Balancer service and select a public IP that CloudStack accepts for LB rules.
3. Create AutoScale at CMP
Customer path: typically Create → Autoscale / VM Autoscale (label may vary by portal theme)
In CMP, creating AutoScale maps to the CloudStack AutoScale model:
| CMP / CloudStack concept | Purpose |
|---|---|
| VM profile | Service offering + template used when launching scaled VMs |
| Scale-up / scale-down policies | Conditions and thresholds (for example CPU above X% for Y duration) |
| VM group | Min / max members, polling interval, link to load balancer rule and profile |
When scale-up fires, CloudStack launches VMs from the template and adds them to the LB. When scale-down fires, excess VMs are removed (subject to min members and grace / expunge behaviour in CloudStack).
Exact CMP Create Autoscale form fields and screenshots can be expanded here when provided. Until then, use the packages and considerations pages above, and validate against your CMP version UI.
4. Billing after create
| Resource | Billing source |
|---|---|
| Auto Scaling engine (optional) | VM Autoscale package |
| Running / scaled VMs | Virtual Machine packages |
| Load balancer | Load Balancer packages |
| Public IPs | IP Address packages |
Validation checklist
Before go-live with AutoScale for a customer:
- App tested with ≥ 2 VMs behind LB
- Golden template boots and serves traffic unattended
- LB uses an eligible public IP (not VPC Source NAT)
- VM Autoscale package is Active for the zone
- Scale-up / scale-down policies and min/max members are intentional
- Quotas allow expected peak VM count