KiviDB Cloud Now Runs on Azure and Google Cloud

Murtaza Shajapurwala ·

KiviDB Cloud now runs on Microsoft Azure and Google Cloud, alongside AWS. Pick the cloud your application already lives in, and your database sits next to it instead of across the internet.

Everything else stays the same: the console, the API, the Terraform provider, the MCP server, and the shape of your hostname.

What's available

AzureGoogle Cloud
Essentials (single node)YesYes
Pro (1 to 3 replicas, automatic failover)Yes, from 8 GBYes, from 8 GB
RegionsCentral India, Sweden Central35 regions across the Americas, Europe, Asia Pacific, the Middle East and Africa
Snapshots and scheduled backupsYesYes
Private connectivityAzure Private LinkPrivate Service Connect
EngineKiviDB 1.0.5 by defaultKiviDB 1.0.5 by default

The free tier and the Scale tier stay on AWS for now.

The Google Cloud guide in the KiviDB docs, with the list of regions you can order in.

Create one

Open the configurator, choose Azure or GCP, and work through the same steps as on AWS: region, memory, throughput, engine version and add-ons. The price updates as you go and is shown before you order.

Your instance gets a hostname like your-db.cloud.kividb.io. It's the same on every cloud, so the hostname doesn't reveal where you run.

An Azure instance in the KiviDB console, with its endpoint, region and plan.

High availability

Pro runs a primary and one to three read replicas. If the primary stops responding, KiviDB promotes a replica and repoints the endpoint, and your application keeps using the same hostname. Replicas go in separate availability zones where the region has them, and the configurator tells you before you order when a region has only one usable zone.

Pro starts at 8 GB on Azure and Google Cloud, compared with 4 GB on AWS. That's a hardware limit, not a pricing choice: the smallest sustained-performance machines on both clouds have two vCPUs.

A Pro instance on Azure with a read replica and its read-only endpoint.

Your data stays on its own disk

Persistence is always on. Every write is appended to a log on disk as it happens, and the disk is separate from the machine: a Premium SSD managed disk on Azure, a Persistent Disk on Google Cloud. If a machine is lost, a replacement starts in front of the same disk with the same data. Snapshots are point-in-time copies you can take on demand or on a schedule, and restore into a new instance.

Private connectivity

Reach your database from your own network without touching the public internet:

  • Azure: KiviDB publishes a Private Link Service, and you create a Private Endpoint in your VNet.
  • Google Cloud: KiviDB publishes a service attachment, and you create a Private Service Connect endpoint in your VPC.

Either way there's no IP allow-list to maintain, and once it's working you can turn public access off entirely. It's $25 per instance per month plus $0.01 per GB, prorated hourly.

When a region is busy

Clouds sometimes run out of a machine type in a region. On Google Cloud, if the machine family your plan maps to is out of stock across your region, KiviDB places your instance on an equivalent family of the same shape. If a cloud has no capacity at all, we tell you plainly, naming the cloud and the region. It's a temporary limit on the provider's side, not on your account, and you can pick another region or try again later.

Same tools, any cloud

The Terraform provider and the MCP server work across all three clouds. One configuration can put a database on each cloud, and an assistant can launch a Pro database on Google Cloud as easily as on AWS:

resource "kividb_instance" "eu" {
  name          = "orders-eu"
  tier          = "pro"
  cloud         = "azure"
  region        = "swedencentral"
  data_size_gb  = 8
  replica_count = 1
}

Get started