Quotas
Quotas are per-project limits that bound how much of any one resource a project can hold. They prevent runaway spend from accidental loops, and they make sure one project's usage can't drain a region's capacity.
What's quotaed
Every project has a quota on each of:
| Resource | Default quota | Notes |
|---|---|---|
| VMs | 50 | Per region; "VM" counts every state except terminated |
| vCPU | 200 | Total across every running VM |
| RAM (GB) | 800 | Total across every running VM |
| Block storage (GB) | 5,000 | Sum of every volume (attached + detached) |
| Object storage (GB) | unlimited | Soft-only: we'll contact you above ~50 TB to align expectations |
| Snapshots | 200 | Per region; deleted snapshots don't count |
| Backups | 500 | Per region |
| Floating IPs | 20 | Per region |
| Networks | 10 | Per region |
| Security groups | 50 | Per region |
| Load balancers | 10 | Per region |
| DNS zones | 50 | Project-wide |
| Users | 25 | Project-wide |
Defaults are conservative. Most production projects get bumped to 5–10× on the most-used axes (vCPU, RAM, storage) shortly after launch.
Quota vs usage
The endpoints expose both:
GET /v1/quotas— the limits (50 VMs, 200 vCPU, etc.)GET /v1/quotas/usage— the current consumption (12 VMs used of 50, 48 vCPU of 200, etc.)
Always check usage before bulk-creating; a request that would push you past quota returns 429 with Retry-After blank (since the quota isn't time-bounded, retrying doesn't help — you need to delete something or raise the quota).
Raising a quota
Quota raises go through the support contact form, not the API. Three reasons:
- Capacity planning. Large raises (10×+) need to reserve hardware in the target region.
- Compliance. Some quota raises require a brief conversation about the workload shape.
- Abuse prevention. A self-service quota-raise endpoint would be the first thing a compromised account would call.
Most raises clear in under a business day. Production migrations get same-day turnaround if you flag them in the request.
Quota across regions
Most quotas are per-region (VMs, snapshots, networks). A few are project-wide (DNS zones, users). The endpoint response makes this explicit — each quota entry carries a scope field (region: us-east-1 or project).
What's next
- Projects — the container quotas are scoped to.
- Authentication — how to call these endpoints.
- The support contact form at /contact/support — where quota raises live.