Costs & billing

Money is the question everyone has, so let's be clear about it. The short version: The Deployer builds real resources in your own cloud account, and your cloud provider bills you directly for them — not us. Here's exactly how it works and how to keep the bill small.

You pay your cloud provider directly

When you deploy, The Deployer creates real, billable resources — servers, storage, databases, and so on — inside your own cloud account. That means your cloud provider bills you directly for what's running. The Deployer never charges you for the resources and never sits between you and your provider's invoice; you're using capacity you rented in your own name.

The estimate is a guide, not the bill

Before and during a deploy, The Deployer shows a cost estimate with a breakdown of the pieces (for example, the server size and any add-ons). This is there to help you plan — but it's an estimate, not a spending cap. The Deployer doesn't stop or limit spending in your account. Your real charges come from your cloud provider and depend on actual usage, your region, data transfer, and your provider's current prices. Always treat the in-app number as a ballpark, and check your provider's own billing dashboard for the exact amount.

What can make the bill higher than the estimate

• Data transfer (egress). Traffic leaving the cloud is billed per GB and grows with your visitors.
• Autoscaling. Architectures that scale out (containers, serverless, Kubernetes) add capacity, and cost, under load.
• Log ingestion and retention. Provider logging and monitoring services bill for the volume they ingest and keep.
• NAT gateways and networking. NAT gateways, load balancers and static IPs have hourly charges plus per-GB processing.
• Managed databases and caches. Their price depends on instance size, storage, backups and I/O, and is often the largest line.

What stopping a server saves

Stopping a server does not stop every charge, and it depends on the cloud:

• AWS, Google Cloud and Azure. A stopped server stops billing for its computing time. Its disk keeps billing, and so does anything stored with it. AWS and Google Cloud also release the server's temporary public address while it is stopped (it changes when the server starts again); on Azure the public address keeps billing.
• DigitalOcean. A powered-off server is billed at the full rate, exactly as if it were running. The only way to stop paying there is to delete it.
• Managed databases. On AWS, Google Cloud and Azure a stopped database stops billing for its instance and keeps billing for its storage. A DigitalOcean database cannot be stopped.

The Cost card shows this as it happens: the price right now with the stopped parts taken out, the full price next to it, and a note on what still bills. The "so far this month" figure is worked out hour by hour from the price that was running in each hour, so a stop, a start and a change of size are all counted for exactly the time they lasted. It is still our arithmetic on estimates, never your cloud bill.

A standby copy of your database (high availability)

By default a managed database is one copy in one place. A standby copy (high availability) is a second copy in another zone of the same region that takes over by itself if the first one fails. It is your choice, it is off unless you turn it on, and you pay for both copies: about twice the price of the database and its storage. The price on the plan and in the Scale dialog changes the moment you switch it, so you see the cost before you agree to it.

Choose it on the plan before you launch, or with Change on the dashboard's Scaling card once the database exists (see Scaling). Turning it on or off for a running database changes the database in your own cloud account: AWS builds the second copy in the background and the database stays up; Google Cloud restarts the database once (expect a few minutes offline); Azure builds the standby server and the database can restart briefly; DigitalOcean adds a standby node with a brief connection blip.

What it covers: the database's server failing, and the provider's maintenance restarts being smoother. What it does not cover:

• A mistake in your data, such as a bad delete. It reaches both copies at once. Backups are what bring data back.
• Your app. Its servers or containers are a separate choice, and a failover can drop connections for a moment, so the app has to reconnect.
• A whole region going down. Both copies are in the same region (on DigitalOcean the standby is a second node in the same data center).
• Speed. The second copy waits; it does not answer queries.

Two clouds have limits worth knowing. Azure only gives a standby copy to the Large size and bigger, because its smaller (Burstable) servers have none. On Google Cloud, the small size has a standby copy but Google's uptime guarantee for high availability covers only instances with a dedicated CPU.

Your real AWS spend (optional)

On an AWS connection you can ask The Deployer to show what AWS says your account has been charged so far this month, next to our estimate. It is off until you turn it on for that connection on the Setup page, because AWS bills you about one cent each time Cost Explorer is read (at most once every 3 hours while you have a dashboard open). The figure is for your whole AWS account, not one project, before credits and refunds, and it can run up to a day behind. Every other number in the Cost card is our estimate.

It needs three things on your side: Cost Explorer switched on in your AWS account (the first time it can take up to 24 hours), the account setting "IAM user and role access to Billing information" turned on by the account owner, and permission for ce:GetCostAndUsage, which the role The Deployer creates already includes (a connection made with your own IAM user needs it added). If one is missing, the card says so in plain words and keeps showing the estimate. DigitalOcean's account figure is free to read and is always shown.

Budgets are alerts, not caps

AWS Budgets, Google Cloud budgets, Azure cost alerts and DigitalOcean billing alerts notify you when spending crosses a threshold. They don't stop spending. Set one up in your provider's billing console as an early warning, and set the threshold below what you're willing to pay.

Deleting a project stops the charges

Because the charges come from running resources, the way to stop paying is to remove them. Deleting a project tears down the resources it created, which is how you avoid paying for something you no longer use. If you spun up a project to try things out and you're done with it, delete it so it isn't quietly costing you money. Removing the resources can take a few minutes on some clouds. The project shows as deleting until they are gone, then it leaves your list.

The Deployer's own plans

Your cloud bill and your Deployer plan are separate. A plan changes what the platform does for you, never what is already running:

• Free: one deployed project at its provider address, manual redeploys, and one automated recovery session per deployment. Scans, verdicts and priced architectures are unlimited.
• Pro: up to five deployed projects, two seats (you and one teammate), custom domains with HTTPS, push-to-deploy, extended automated recovery, and uptime and deploy alerts.
• Team: everything in Pro for up to fifteen deployed projects and ten seats, with priority support. See working with your team.

A repository with several services, such as a website and an API, is deployed as one project per service. Projects from the same repository and branch count as one deployed project.

Current prices are on the Plans & Billing page. If you stop paying, your apps keep running in your own cloud account and your account moves to the Free plan. Cancelling a plan doesn't delete any cloud resources, so they keep billing in your provider account until you delete the project or remove them yourself.

Tips to keep costs low

• Start small. Choose a smaller server size to begin with — you can always scale up later if the app needs it. Small sizes are dramatically cheaper.
• Delete projects you're not using. An idle deploy still bills. Tearing it down is the single most effective way to cut costs. Stopping a server saves its computing time on AWS, Google Cloud and Azure, but not its disk, and nothing on DigitalOcean.
• Keep an eye on the Cost card. Check the estimate before you launch, and re-check it when you change sizes or add services.
• Mind your provider's own pricing. Things like data transfer (traffic leaving the cloud) and managed databases add up. Your provider's billing page shows where the money actually goes.

Ready to deploy

The quick start walks you through a full deployment, and cloud provider limits & quotas covers the account rules that can affect it.

Paying in rupees

In India, Pro and Team can be paid in rupees by UPI or card. Choose INR (UPI) above the plans. The rupee price includes 18% GST. A plan you already pay for in dollars renews in dollars until its term ends; after that you can switch.

Partner codes

If a creator or agency gave you a code, type it at checkout. It takes its discount off your first purchase, a plan or a done-for-you order. If you signed up through their link, the discount is already applied.