Azimutt on GKE Autopilot

Azimutt is an open-source, next-generation database-schema explorer and ERD (entity relationship diagram) tool for real-world databases, built with Elixir/Phoenix. It lets teams explore, document, and design large schemas (thousands of tables), search across columns and relations, and share diagrams. This module deploys Azimutt on GKE Autopilot on top of the App_GKE foundation, which provisions and manages the shared Google Cloud and Kubernetes infrastructure.

This guide focuses on the cloud services Azimutt uses and how to explore and operate them from the Google Cloud Console and the command line. For the mechanics that are common to every GKE application — Workload Identity, ingress, autoscaling, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC Service Controls, backups, and the deployment lifecycle — refer to the App_GKE foundation guide rather than repeating them here.

What Azimutt costs on RAD, and how that compares

Azimutt on RAD's GKE Autopilot module costs about US$136 a month in a project you own, with no licence fee for the software itself. RAD charges a one-off module fee of 175 credits (US$17.50 at the top-up price) in a project you own, and 157.5 credits (10% lower) in a project RAD manages. GKE suits Azimutt that must stay up continuously or scale across pods; for the lowest cost, the Cloud Run module runs the same Azimutt for about US$89 a month. Figures are as at 8 October 2026; sources are listed at the end of this section.

If you only need it occasionally — studying for a certification, a demo, a seasonal need — a RAD-managed project can be deleted and restored within 30 days for a handful of credits, so that about US$136 a month becomes close to nothing instead. See Pause it for free, below.

What you pay on RAD

You pay for three things:

	In a Google Cloud project you own	In a project RAD manages for you
Module fee, once per deployment	175 credits (US$17.50 at the top-up price)	157.5 credits (10% lower)
Build time	A few credits, metered per minute of build time	The same
Google Cloud running cost	Billed by Google to your own billing account, about US$136 a month for the default configuration (table below)	Metered hourly in credits; RAD publishes 105 credits a day, about 3,150 a month (about US$315 at the top-up price, US$252 at the Scale plan's rate)
Who owns what	You own the project and its billing; RAD deploys and updates it	RAD owns the project, with guardrails, quotas and budget alerts; you get console access, and a minimum purchased balance is held in reserve

10 credits cost US$1 on a one-off top-up, and less on a monthly plan. In a project RAD manages, the database is shared by every application in the project, so a second application does not add a second one.

Default running cost in your own project (us-central1, Google list prices):

Resource (module default)	Per month
GKE Autopilot pod, 2x vCPU / 4 GiB	US$79.35
Cloud SQL for PostgreSQL, 1 vCPU / 3.75 GB, zonal, 10 GB SSD	US$51.02
Cloud NAT and networking	US$5
Cloud Storage (add-ons, backups)	US$1
GKE cluster management fee	US$0 if this is your only Autopilot/zonal cluster on the billing account, otherwise about US$73 (shared across every GKE app in the project)
Total	about US$136

How it compares

No name-brand SaaS competitor is a clean match for Azimutt, so the honest comparison is a plain self-managed server. Running the same open-source software yourself, a Hetzner CPX22 (2 vCPU/4 GB, about $24/month) or a DigitalOcean 2 vCPU/4 GB Droplet (also $24/month) is roughly the same shape; a GCP Compute Engine e2-standard-2 (2 vCPU/8 GB) runs about $49/month.
RAD is not cheaper than a bare VPS in cash terms, and does not try to be. What RAD adds for a similar price is a managed database, Secret Manager for credentials, and monitoring that would otherwise be your own job to set up, patch and keep backed up.
The admin's own time is the real cost a VPS does not show. Security patches, backups and version upgrades on a bare server are work someone has to keep doing; RAD's managed services absorb that.

Pause it for free: delete a RAD-managed project, restore it when you need it

If Azimutt runs in a project RAD manages for you, you have an option that goes well beyond scaling to zero: delete the whole project, and restore it within 30 days for close to nothing. This suits occasional use — studying for a certification, a demo, a seasonal need — far better than running Azimutt continuously.

How it works. Deleting a RAD-managed project unlinks its billing first, then asks Google to delete the project outright. Google does not remove it immediately: it keeps the project, recoverable, for 30 days, and because billing is already unlinked nothing is charged while it waits. Unlike deleting one module, this does not tear down the database, any VM or the compute resource one by one — the whole project simply stops, and nothing is billed in the meantime.
Restoring costs a handful of credits, not a rebuild. Within 30 days, only the project's owner can restore it. RAD asks Google to undelete the project and reattaches its billing account, then asks you to run Update on each deployment to confirm everything came back. Because nothing was individually destroyed, that Update finds the same resources already there — it is a check, not a rebuild, and an Update never charges the module fee again. This costs only a handful of credits in total (under US$1), against paying the module fee and a full build again.
What this needs. You must own the project (not one RAD only manages billing for), you must restore it yourself within the 30 days — after that Google deletes it for good — and restoring is admitted like creating a new project, so your purchased credit balance must still clear the tier's floor (100 credits for the sandbox tier most study and demo use fits). Google says most services are fully working again within 36 hours of a restore.
One real gap: backups do not survive. Anything Azimutt writes to a backup bucket inside the project is very likely gone as soon as you delete the project, even though the project itself is recoverable for 30 days — that bucket has Cloud Storage's soft-delete explicitly turned off. Anyone who has customised the deployment and wants to keep that work should copy a backup out (to Google Drive, or a bucket outside the project) before deleting. For a default install with nothing irreplaceable in it, this does not matter.

Pay only while you use it, the other way: delete and redeploy

This is the fallback for a project you own, or once the 30-day restore window above has passed: delete the deployment, and deploy it again when you next need it.

What a redeploy costs. The module fee again, plus the builds: roughly 180 credits (about US$18) in your own project, or 162.5 credits in a RAD-managed one, because RAD rebuilds the resources from scratch. Deleting saves money once Azimutt would otherwise sit unused long enough to clear that redeploy cost against its own running cost — roughly 4 days or more in your own project (US$4.55/day) or 2 days or more in a RAD-managed one (105 credits/day).
Most of the running cost is usually shared infrastructure. The database stops only when nothing else in the project uses them, so deleting Azimutt while another application shares the project saves only Azimutt's own compute part.
Keep your data first. Anything Azimutt writes to a backup bucket inside the deployment is deleted with it — copy the latest backup out (to Google Drive, or a bucket you keep) before deleting if you want to keep it.
RAD does not recreate a deleted deployment for you. You enter the settings again when you redeploy.

Lab sessions and Managed Environments

Lab sessions, for training. A trainer runs a session for a class. Each participant gets Azimutt in their own Google Cloud project for 15 minutes to 24 hours, within an allowance the trainer sets. Either the trainer funds every place, or each participant pays for their own. Everything is deleted when the session ends and unused credits go back to the trainer.
Managed Environments, for consultancies. A partner runs Azimutt for a client from a ring-fenced wallet it funds, and settles with the client directly. If the wallet runs low, billing pauses and the data is kept, so nobody receives an unexpected charge. At the end the partner hands the project over and the deployments become the client's own.

Credits can be bought in more than 20 currencies, including XAF, XOF, NGN, GHS, KES and ZAR, by card, bank transfer or mobile money.

Sources (8 October 2026): Google Cloud list prices from the Cloud Billing Catalog API; RAD's fees and daily-credit estimate from radmodules.dev/pricing; Delete and restore projects for the pause/restore mechanics. Prices change; check each source before relying on a figure.

1. Overview

Azimutt runs as a single Elixir/Phoenix web workload listening on port 4000. The deployment wires together a focused set of Google Cloud services:

Capability	Google Cloud service	Notes
Compute	GKE Autopilot	Phoenix pods, horizontally autoscaled; bills for requested CPU/memory
Database	Cloud SQL for PostgreSQL 15	Required — Azimutt does not support MySQL or other engines
File storage	Cloud Filestore (NFS)	Optional, off by default (`enable_nfs = false`): Azimutt writes uploads to its own ephemeral working directory, never to this mount (see below)
Object storage	Cloud Storage	A bucket is provisioned (available for an S3-compatible file adapter)
Secrets	Secret Manager	Auto-generated Phoenix `SECRET_KEY_BASE`; database password
Image build	Cloud Build + Artifact Registry	Thin wrapper FROM `ghcr.io/azimuttapp/azimutt`, mirrored into Artifact Registry
Ingress	Cloud Load Balancing	External LoadBalancer, optional custom domain + managed certificate

Sensible defaults worth knowing up front:

PostgreSQL 15 is mandatory. The database engine is fixed by the shared application layer; selecting any other engine breaks startup. All Azimutt project data lives in Postgres.
Azimutt connects to Postgres over the Auth Proxy loopback (no SSL). Ecto cannot parse the Cloud SQL socket DSN, so on GKE the entrypoint builds DATABASE_URL against 127.0.0.1 (the Cloud SQL Auth Proxy sidecar, which terminates TLS) with DATABASE_ENABLE_SSL=false. enable_cloudsql_volume = true is required.
container_port and probes must be 4000. On GKE the platform does not auto-inject PORT, so the entrypoint defaults PORT=4000; the Service port and probes must match or the pod never becomes Ready even though the app is healthy.
NFS is off by default (enable_nfs = false) because nothing in this module writes to it — FILE_STORAGE_ADAPTER stays local, so uploads land on the pod's ephemeral disk whether or not NFS is mounted. Project data itself (schemas, diagrams, layouts, users) lives in Postgres and is unaffected.
SECRET_KEY_BASE is generated automatically and stored in Secret Manager. Rotating it after first boot signs out every active session; only rotate in a maintenance window.
Minimum 1 replica is maintained (min_instance_count = 1) — GKE does not support scale-to-zero, keeping Azimutt always reachable.
Migrations run automatically on every boot (/app/bin/migrate && /app/bin/server); allow extra time on the first boot.
application_version = "latest" maps to Azimutt's main tag. Pin to a specific release in production.
Sign-up is open by default. Restrict access after creating your first account.

2. Google Cloud Services & How to Explore Them

All commands assume you have run gcloud container clusters get-credentials <cluster> --region <region> --project <project> and that PROJECT, REGION, and NAMESPACE are set. The namespace and other identifiers are reported in the deployment Outputs.

A. GKE Autopilot — the Azimutt workload

Azimutt pods are scheduled on Autopilot, which bills for the CPU/memory the pods actually request. Horizontal Pod Autoscaling sizes the deployment between the minimum and maximum replica counts.

Console: Kubernetes Engine → Workloads → select the Azimutt workload to see pods, revisions, and events. Kubernetes Engine → Services & Ingress shows the external IP.

CLI:

kubectl get pods,svc,hpa -n "$NAMESPACE"
kubectl logs -n "$NAMESPACE" deploy/<service-name> --tail=100
kubectl logs -n "$NAMESPACE" deploy/<service-name> | grep cloud-entrypoint  # resolved DB wiring

See App_GKE for how Autopilot, scaling, and the workload type (Deployment vs StatefulSet) are managed.

B. Cloud SQL for PostgreSQL 15

Azimutt stores all application data (schemas, diagrams, layouts, users, sources) in a managed Cloud SQL for PostgreSQL 15 instance. Pods reach it through the Cloud SQL Auth Proxy sidecar on 127.0.0.1 (TLS terminated by the proxy, so DATABASE_ENABLE_SSL=false). On first deploy an initialization Job creates the application database and role; Azimutt then runs its own Ecto migrations on boot.

Console: SQL → select the instance for connections, backups, flags, metrics.

CLI:

gcloud sql instances list --project "$PROJECT"
gcloud sql instances describe <instance-name> --project "$PROJECT"
gcloud sql connect <instance-name> --user=<db-user> --database=<db-name> --project "$PROJECT"

The instance name, database, user, and password secret are all surfaced in the Outputs. For the connection model, backups, and password rotation, see App_GKE.

C. Cloud Filestore (NFS) & Cloud Storage

NFS is off by default (enable_nfs = false); if enabled it is mounted at nfs_mount_path, but it is not wired to Azimutt's storage path — with the default FILE_STORAGE_ADAPTER = local, Azimutt still writes uploads to its own ephemeral working directory rather than this mount. A Cloud Storage bucket is also provisioned (available if you switch Azimutt to an S3-compatible file adapter). Project data itself (schemas, diagrams, layouts, users) lives in Postgres and is unaffected either way.

Console: Filestore → Instances; Cloud Storage → Buckets.

CLI:

gcloud filestore instances list --project "$PROJECT"
gcloud storage buckets list --project "$PROJECT"
kubectl get pvc -n "$NAMESPACE"

See App_GKE for NFS provisioning, CMEK options, and GCS Fuse mounts.

D. Secret Manager

The Phoenix SECRET_KEY_BASE is generated automatically and stored in Secret Manager (used to sign and encrypt session cookies). The database password is managed separately by the foundation.

Console: Security → Secret Manager.

CLI:

gcloud secrets list --project "$PROJECT" --filter="name~secret-key-base"
gcloud secrets versions access latest --secret=<secret-name> --project "$PROJECT"

See App_GKE for the Secret Store CSI integration and rotation.

E. Cloud Build & Artifact Registry

Azimutt's image is a thin wrapper built FROM ghcr.io/azimuttapp/azimutt; Cloud Build produces the wrapped image and it is mirrored into Artifact Registry (enable_image_mirroring = true). Because it is a rebuilt/mirrored image, App_GKE sets imagePullPolicy = Always so nodes never serve a stale cached layer.

Console: Cloud Build → History; Artifact Registry → Repositories.

CLI:

gcloud builds list --project "$PROJECT" --limit 5
gcloud artifacts docker images list <region>-docker.pkg.dev/$PROJECT/<repo> --include-tags

F. Networking & ingress

By default the workload is exposed through an external Cloud Load Balancing IP (service_type = LoadBalancer). A custom domain with a Google-managed certificate can be enabled, and a static IP can be reserved so the address survives redeploys. session_affinity = ClientIP keeps a client pinned to one pod.

Console: Network services → Load balancing; VPC network → IP addresses.

CLI:

kubectl get ingress,svc -n "$NAMESPACE"
gcloud compute addresses list --project "$PROJECT"

See App_GKE for custom domains, Cloud CDN, and static IP details.

G. Cloud Logging & Monitoring

Pod stdout/stderr flow to Cloud Logging; GKE and Cloud SQL metrics flow to Cloud Monitoring. Optional uptime checks and alert policies are available. The cloud-entrypoint lines show the resolved DATABASE_URL path, PHX_HOST, and PORT.

Console: Logging → Logs Explorer; Monitoring → Dashboards / Alerting.

CLI:

gcloud logging read 'resource.type="k8s_container" AND resource.labels.namespace_name="'"$NAMESPACE"'"' \
  --project "$PROJECT" --limit 50

3. Azimutt Application Behaviour

First-deploy database setup. An initialization Job runs db-init.sh using postgres:15-alpine. It idempotently creates the application role (LOGIN CREATEDB) and database, grants ALL on the database and the public schema, and ALTERs the schema owner — Azimutt needs full DDL rights because it runs its own migrations. The job then signals the Auth Proxy sidecar to shut down (/quitquitquit) so the Job pod completes. Safe to re-run.
Migrations run on start. The container command is /app/bin/migrate && /app/bin/server, so Ecto applies pending migrations on every boot before the Phoenix endpoint binds. Upgrading application_version applies schema changes automatically.
Runtime DB wiring is composed by the entrypoint. On GKE DATABASE_URL is built against the Auth Proxy loopback (127.0.0.1) with DATABASE_ENABLE_SSL=false, and PORT is defaulted to 4000 (GKE does not auto-inject it). PHX_HOST is derived from the injected service URL.
SECRET_KEY_BASE is stable and effectively immutable. Rotating it invalidates every active session cookie — all users are signed out. Only rotate in a maintenance window.
Health path. The startup probe targets the Phoenix root / and the liveness probe /health, both with a 60-second initial delay. The liveness probe is mirrored into the Gateway health check, which requires a literal 200, so it must not point at /. The probes and container_port must both be 4000 or the pod never becomes Ready.
First-run setup. Reach the service via its external LoadBalancer IP (or custom domain) and create the first Azimutt account through the sign-up page. Sign-up is open by default — restrict access afterwards.

Inspect the init job execution:

kubectl get jobs -n "$NAMESPACE"
kubectl logs -n "$NAMESPACE" job/<job-name>

4. Configuration Variables

Variables are grouped exactly as they appear on the deployment platform. Only settings specific to or notable for Azimutt are listed; every other input is inherited from App_GKE with its standard behaviour and defaults.