Deploy
Your cloud, any cloud, or your own datacenter
SOMVAD ships as native, CNCF-conformant Kubernetes workloads with Terraform and Helm. The same artifacts run on any conformant cluster, so you are never locked to one cloud, and you can keep everything inside your own boundary when you need to.
One command per region
Terraform plans it, Helm ships it
Infrastructure is code: a Terraform module provisions the cluster and network, a Helm chart deploys the workloads. GKE is wired today; because the workloads are standard Kubernetes objects, the same chart targets any conformant cluster with no per-cloud rewrite.
- VPC-native cluster, node pools, firewall, static IPs, DNS, secrets
- Helm chart: migration job, control plane, media mixers, Redis, gateway, TURN, ingress + managed certs
- Autoscaling media pool; cert-manager for TLS
# GKE is wired today $ make deploy gcp:europe-west2 # the same Terraform + Helm pattern, per target $ make deploy aws:eu-west-1 $ make deploy azure:uksouth $ make deploy onprem:dc-1 # under the hood, every target terraform apply + helm upgrade --install somvad
Illustrative of the deploy interface. The reference one-command pipeline ships for GKE; the same chart deploys to other conformant clusters.
Why it ports cleanly
Everything is a CNCF-native workload
No proprietary cloud primitive sits on the critical path. If a cluster is CNCF-conformant, the stack runs on it.
Standard objects
Deployments, StatefulSets, Services, Ingress, Jobs, ConfigMaps, Secrets, PDBs, HPAs. Nothing exotic.
Portable building blocks
Kubernetes, Helm, cert-manager, Terraform, FreeSWITCH, Redis, Postgres. Open, swappable, no single-cloud lock-in.
Pluggable edges
Blob storage (GCS / S3 / local), identity (OIDC), and DB are interfaces, so each maps to the equivalent on your cloud.
Sovereign by deployment
On-prem & air-gapped
Run the whole platform inside your own datacenter or private cloud. Media, recordings, and identity stay within your boundary, which is what regulated and sovereignty-sensitive operations need. No call audio or recording leaves your infrastructure.
Bring your own carrier trunks, your own object storage, and your own IdP. The control plane is the same one we run.
| Concern | How it maps |
|---|---|
| Database | Postgres (Cloud SQL, RDS, or self-managed) |
| Object storage | GCS / S3 / on-prem (BlobStore interface) |
| Identity | Any OIDC IdP (Entra, Keycloak, Okta…) |
| TLS | cert-manager (Let's Encrypt or internal CA) |
| Carrier (PSTN) | Your SIP trunks via the gateway |
Today's reference deployment and one-command pipeline ship for GKE; because the workloads are standard CNCF objects and the cloud edges are interfaces, the same Helm chart targets other conformant clusters. Talk to us about your target cloud or on-prem environment.