Concepts

Private cloud

Physical hosts dedicated to one account, how to request them, the minimums, how a request moves on and the overview of your private capacity.

A private cloud is a set of physical hosts Ankra dedicates to your account. No other customer's server is placed on them, and your servers in a zone with private hosts run only there, never on shared capacity. Everything else stays the same: you deploy and manage servers with the same console, API, CLI, Terraform provider and MCP tools as on the public cloud.

Requesting one#

In the console, open Private cloud and fill in the request: the vCPUs, memory and disk you need, optionally the region the hosts should stand in, and notes for Ankra (workloads, timing, who to contact). The same works over the API:

bash

curl -X POST https://cloud.ankra.app/v1/private-cloud/requests \
  -H "Authorization: Bearer $ANKRA_CLOUD_TOKEN" \
  -H 'Content-Type: application/json' \
  -d '{ "vcpus": 32, "memory_gibibytes": 128, "disk_gigabytes": 2000, "region": "eu-north", "notes": "CI runners from November" }'

A request starts at 16 vCPUs, 64 GiB of memory and 500 GB of disk by default; GET /v1/private-cloud returns the current minimums and maximums. Requesting needs the billing.manage permission (account owners and admins, or an API token granted it). Nothing is provisioned or billed by the request itself: Ankra staff contact you first.

How a request moves on#

State	Meaning
`open`	Sent, waiting for Ankra staff.
`in_review`	Ankra staff are working on it.
`fulfilled`	The hosts are dedicated to your account.
`declined`	Ankra could not offer it; the request carries the reason.
`withdrawn`	You took it back.

An account has one open request at a time. Withdraw an open or in_review request with DELETE /v1/private-cloud/requests/{id} (or Withdraw in the console) before sending another. Every request and withdrawal is recorded in your audit log.

The overview#

Once hosts are dedicated to your account, Private cloud shows them as capacity, never as machines:

the totals across your private hosts: how many hosts, the vCPU budget and memory with what your servers use, and the hosts' disk;
each zone's dedicated capacity, used and free;
the servers placed on your private hosts, each linking to its page.

The vCPU budget counts four vCPUs per logical CPU, the way servers are placed; each host keeps 2 GiB of memory for itself. Request more capacity sends a new request with the same form. The API behind the page is get_private_cloud.