# Dedicated deployments

Source: https://docs.littlebigbrain.com/guides/pro-deployments/

A dedicated deployment is part of Enterprise. Free, Cloud and Pro run in the
shared cloud, and Pro includes every feature there.
[Talk to us](mailto:hello@littlebigbrain.com?subject=Enterprise%20deployment)
to enable dedicated deployments for your workspace. After that, **Deployment**
shows in the sidebar. Enterprise billing runs on a Pro subscription.

The **Deployment** page offers these choices. Hosted EU supports automatic setup
when the hosted provisioner is enabled. AWS, Google Cloud and Azure support
automatic setup through their Cloud Shells when a provisioning release is
enabled:

- **Little Big Brain Cloud**: dedicated compute and storage in Europe. The
  United States option is shown as coming soon and cannot be selected.
- **Your AWS cloud**: enter your 12-digit account ID and AWS region.
- **Your Azure cloud**: enter your subscription ID, Microsoft Entra directory
  (tenant) ID and region, for example `westeurope`. Both IDs are UUIDs.
- **Your Google Cloud**: enter your GCP project ID and region, for example
  `my-lbb-project` and `europe-west1`. Use the project ID, not its display name
  or numeric project number. Use a region, not a zone such as `europe-west1-b`.
  Deployments use native Google Cloud Storage with the runtime's Google service
  identity. The installer checks native storage and database health before reporting readiness.

Do not enter access keys or service-account credentials. US regions can be
requested in any customer cloud; hosted US remains unavailable.

A hosted deployment is managed automatically: Little Big Brain applies
releases and fixes as they are scheduled. An own-cloud deployment chooses **With my approval** to review releases and restarts, or
**Managed automatically**. Select
**Save and continue** to save the setup. This step does not create cloud
resources or start billing. Payment comes right before the launch: the launch
buttons stay disabled until your Pro subscription is active. After checkout,
open **Deployment** again. For your own cloud, select **Launch in AWS**, **Launch in
Google Cloud** or **Launch in Azure**, open the linked Cloud Shell using your
cloud administrator login, run the supplied command and paste the one-hour code
when prompted. Review the infrastructure plan before approving resource
creation. A subscription does not mean your infrastructure has finished
provisioning.

You can change a setup request before launching it or preparing its first
launch code. Open **Deployment** in the sidebar to return to the setup and its
operations.

## Launch in Europe

Save the preselected **Little Big Brain Cloud · Hetzner → Europe**, activate your Pro subscription, then
select **Launch in Europe**. Little Big Brain creates a dedicated starter VM,
private network, load balancer, storage bucket and HTTPS endpoint. No cloud
account or Cloud Shell is needed. You can close the page while setup continues.

The page tracks infrastructure creation, storage/database checks and public
endpoint verification. **Your cloud is ready** appears only after all checks
pass. Copy the endpoint and select **Get API key** to connect with an SDK or the
API. API keys are available only to owners/admins with key-management access.
A cell is dedicated to your workspace and can hold multiple environments. It
starts with Main; existing shared-cloud environments are not moved.

Select **Add environment**, enter a name and an ID such as `staging`, and select
**Create environment**. It uses the existing cell's compute and storage; no new
cell is launched. Select an environment to copy its endpoint and retrieve its
own API key once its HTTPS checks pass. The starter configuration allows five
environments including Main. Failed creation offers **Retry environment**, which
keeps that environment's identity and data.

Environments have separate data, graphs and keys. They share capacity, outages
and cell-wide updates; use separate deployments if you need independent release
or failure boundaries. This console environment-creation flow currently applies
to hosted EU. BYOC installers still create Main initially.
Dedicated data browsing in the console is not yet available.

If setup fails, **Retry deployment** resumes the same cell and preserves its data.
Failed launches retain infrastructure; contact support for removal.
The starter cell uses one VM; multi-node availability is
not part of this initial setup.

## Updates and fixes

A management agent runs in your deployment and connects out to Little Big
Brain's control plane. It fetches approved operations and reports progress.
An upgrade or restart applies to the whole cell and all its environments.
Database queries and writes continue independently if the control plane is
unavailable. Agent connection is shown separately from deployment readiness.

The deployment page shows proposed operations, their purpose and their status.
On an own-cloud deployment with approval required, an owner or admin selects
**Approve upgrade** or
**Approve restart** before the agent can fetch the operation. Members have
read-only access. Contact support if your deployment needs attention.

## Private access and costs

AWS launches a private EC2 instance with native S3 and an instance role. Google
Cloud launches a private GCE VM with native GCS and a service account. The initial
size is 4 vCPU, 16 GB RAM and a 100 GB disk. Azure uses a private VM, native
Blob Storage and managed identity, with a 128 GB disk. Cloud infrastructure charges go to
your cloud account, including the NAT gateway, disk and object storage.

After setup passes its checks, the page gives you a private Session Manager
(AWS) or IAP (Google Cloud) tunnel command and a command to retrieve your API key
from your cloud's secrets manager. Keep the tunnel open and connect your client
to `http://localhost:7400`. For AWS, install the AWS CLI and
[Session Manager plugin](https://docs.aws.amazon.com/systems-manager/latest/userguide/session-manager-working-with-install-plugin.html)
on the computer where you open the tunnel. The hosted console tracks deployment
and updates; it does not route data queries through your private tunnel.

For Azure, choose Bash in Cloud Shell and use an Owner login for the selected
subscription. Keep a persistent Cloud Shell home for Terraform state. After
launch, run the supplied verified connection helper on your **local Mac or Linux
computer**, with Azure CLI (`az login`), its Bastion extension
(`az extension add --name bastion`), Python 3 and OpenSSH. Bastion connections
cannot run in Cloud Shell. The helper retrieves your connection key from Key
Vault, verifies the VM's SSH host key, and forwards `localhost:7400` through
Bastion and SSH. Leave it open while using the database.

Azure also bills Standard Bastion while idle, NAT Gateway, gateway public IPs and
Key Vault. The stop command deallocates the VM and removes Bastion/NAT/IPs, retaining
your disk, Blob versions and secrets. If you cancel the removal plan after
deallocation, the gateways keep accruing charges until you rerun stop. Azure
public cloud is supported; sovereign clouds and multi-node availability are not
part of this initial setup.

The **Stop or resume your test VM** section gives the command for the original
Cloud Shell. Stopping preserves your disk, data, credentials and update journal;
these retained resources remain billed. Keep `~/lbb-deployments/` in Cloud Shell:
it holds the state needed to retry, stop and resume. If setup fails, retain that
state, fix the reported prerequisite and get a new setup code to resume.

A launch message saying no release is published means provisioning has not yet
been enabled. Contact [hello@littlebigbrain.com](mailto:hello@littlebigbrain.com)
for release access or dedicated hosted setup.